WEBVTT 00:00:00.000 --> 00:00:00.510 align:middle line:90% 00:00:00.510 --> 00:00:04.410 align:middle line:84% ATUL BUTTE: So last week I talked about-- 00:00:04.410 --> 00:00:05.430 align:middle line:90% what did I talk about? 00:00:05.430 --> 00:00:09.090 align:middle line:84% I talked about an introduction to molecular biology 00:00:09.090 --> 00:00:10.710 align:middle line:90% for about 10 or 12 slides. 00:00:10.710 --> 00:00:12.660 align:middle line:84% And then we were going pretty quickly 00:00:12.660 --> 00:00:19.170 align:middle line:84% I covered a lot of this material on gene measurement techniques. 00:00:19.170 --> 00:00:22.527 align:middle line:84% And so we've already had one question, or one request, 00:00:22.527 --> 00:00:25.110 align:middle line:84% to talk about something that I actually flipped pretty quickly 00:00:25.110 --> 00:00:28.170 align:middle line:84% through, this concept of SAGE and how is SAGE 00:00:28.170 --> 00:00:29.580 align:middle line:90% used to measure microarrays. 00:00:29.580 --> 00:00:31.573 align:middle line:84% And actually it's a measure of RNA, 00:00:31.573 --> 00:00:33.990 align:middle line:84% and it's quite important that we actually talk about SAGE, 00:00:33.990 --> 00:00:34.870 align:middle line:90% I think. 00:00:34.870 --> 00:00:36.412 align:middle line:84% But I'm going to start the discussion 00:00:36.412 --> 00:00:38.730 align:middle line:84% with SAGE talking about microarrays again for a second. 00:00:38.730 --> 00:00:40.272 align:middle line:84% So I we talked about microarrays, how 00:00:40.272 --> 00:00:42.060 align:middle line:90% these are made grids of DNA. 00:00:42.060 --> 00:00:45.930 align:middle line:84% Each spot in the grid is looking for one particular RNA. 00:00:45.930 --> 00:00:48.570 align:middle line:84% And in fact, more than one spot might be actually hybridized 00:00:48.570 --> 00:00:49.920 align:middle line:90% into one RNA. 00:00:49.920 --> 00:00:52.297 align:middle line:84% It doesn't have to be one to one It could be many to one. 00:00:52.297 --> 00:00:54.630 align:middle line:84% And I just want to talk about the differences of the two 00:00:54.630 --> 00:00:56.400 align:middle line:84% technologies for microarrays and then 00:00:56.400 --> 00:00:59.100 align:middle line:84% talk about how that actually leads itself 00:00:59.100 --> 00:01:01.750 align:middle line:90% to the SAGE discussion. 00:01:01.750 --> 00:01:04.709 align:middle line:84% So for the past coming up on 10 years now 00:01:04.709 --> 00:01:08.100 align:middle line:84% there have been two technologies to make microarrays. 00:01:08.100 --> 00:01:10.890 align:middle line:84% There's the oligonucleotide approach and the cDNA approach. 00:01:10.890 --> 00:01:12.840 align:middle line:84% The analogy I like to use is this. 00:01:12.840 --> 00:01:16.320 align:middle line:84% If you think of a microarray as, say, Manhattan with all the 00:01:16.320 --> 00:01:18.030 align:middle line:90% skyscrapers. 00:01:18.030 --> 00:01:19.890 align:middle line:84% If you had to rebuild Manhattan, there's 00:01:19.890 --> 00:01:21.420 align:middle line:90% two ways you could do it. 00:01:21.420 --> 00:01:25.425 align:middle line:84% You could take one skyscraper at a time and put them into place, 00:01:25.425 --> 00:01:27.300 align:middle line:84% or you could just build all the first floors, 00:01:27.300 --> 00:01:29.220 align:middle line:84% all the second floors, all the third floors, 00:01:29.220 --> 00:01:30.277 align:middle line:90% all the fourth floors. 00:01:30.277 --> 00:01:31.860 align:middle line:84% And that's essentially the differences 00:01:31.860 --> 00:01:33.450 align:middle line:90% between these two technologies. 00:01:33.450 --> 00:01:36.270 align:middle line:84% So the oligonucleotide arrays are really made by Affymetrix. 00:01:36.270 --> 00:01:38.700 align:middle line:84% They're made-- each of the skyscrapers-- 00:01:38.700 --> 00:01:40.890 align:middle line:84% the equivalent, each of the strands of DNA, 00:01:40.890 --> 00:01:43.470 align:middle line:84% are only 25 nucleotides long, but they're 00:01:43.470 --> 00:01:45.883 align:middle line:90% built one base pair at a time. 00:01:45.883 --> 00:01:47.550 align:middle line:84% Because they're only 25 nucleotides long 00:01:47.550 --> 00:01:49.110 align:middle line:84% and genes are much longer than that, 00:01:49.110 --> 00:01:51.030 align:middle line:84% that's why they have multiple probes going 00:01:51.030 --> 00:01:52.890 align:middle line:90% against a particular gene. 00:01:52.890 --> 00:01:55.650 align:middle line:84% The cDNA, on the other hand, are essentially 00:01:55.650 --> 00:01:58.650 align:middle line:84% strands of DNA that will hybridize to the RNA, 00:01:58.650 --> 00:02:00.720 align:middle line:84% but the strands of DNA that are just 00:02:00.720 --> 00:02:02.700 align:middle line:84% put into place using primarily using something 00:02:02.700 --> 00:02:04.020 align:middle line:90% like a robotic spotter. 00:02:04.020 --> 00:02:06.390 align:middle line:84% So it might pick up four at a time, or 16 at a time, 00:02:06.390 --> 00:02:09.930 align:middle line:84% or just one at a time, but that's how these two are made. 00:02:09.930 --> 00:02:12.900 align:middle line:84% And beyond the technology difference here 00:02:12.900 --> 00:02:15.960 align:middle line:84% there's a major difference in what exactly 00:02:15.960 --> 00:02:18.360 align:middle line:90% these microarrays are measuring. 00:02:18.360 --> 00:02:22.170 align:middle line:84% The cDNA arrays are measuring relative expression amounts. 00:02:22.170 --> 00:02:26.460 align:middle line:84% So you have to put two samples on there, one of which 00:02:26.460 --> 00:02:28.690 align:middle line:84% the RNA might be colored red, another of which 00:02:28.690 --> 00:02:30.450 align:middle line:90% the RNA might be colored green. 00:02:30.450 --> 00:02:32.520 align:middle line:90% And you'll look at-- 00:02:32.520 --> 00:02:34.920 align:middle line:84% what people will do is after they've hybridized it, 00:02:34.920 --> 00:02:36.782 align:middle line:84% they've taken a TIFF image of this, 00:02:36.782 --> 00:02:39.240 align:middle line:84% they'll look at the relative ratio of how much red is there 00:02:39.240 --> 00:02:40.380 align:middle line:90% compared to how much green. 00:02:40.380 --> 00:02:44.400 align:middle line:84% So in that you always have a relative expression level. 00:02:44.400 --> 00:02:47.310 align:middle line:84% In theory the oligonucleotide approach is different. 00:02:47.310 --> 00:02:50.160 align:middle line:84% Now this is relative, because each of the strands of cDNA 00:02:50.160 --> 00:02:51.960 align:middle line:90% has its own properties. 00:02:51.960 --> 00:02:54.780 align:middle line:84% Each of the strands might have a different set of genes and As 00:02:54.780 --> 00:02:57.160 align:middle line:84% and Ts and Cs, different melting temperatures. 00:02:57.160 --> 00:02:58.827 align:middle line:84% And that's why you can't really use that 00:02:58.827 --> 00:03:00.090 align:middle line:90% as an absolute measurement. 00:03:00.090 --> 00:03:03.720 align:middle line:84% And intensity might be lower on one of these arrays, 00:03:03.720 --> 00:03:05.610 align:middle line:84% not because there's less of a gene present, 00:03:05.610 --> 00:03:08.505 align:middle line:84% just only because the actual cDNA had 00:03:08.505 --> 00:03:10.380 align:middle line:84% a little bit different biochemical properties 00:03:10.380 --> 00:03:13.740 align:middle line:84% and wasn't hybridizing efficiently. 00:03:13.740 --> 00:03:15.670 align:middle line:84% The oligonucleotide arrays on the other hand, 00:03:15.670 --> 00:03:18.810 align:middle line:84% are, in theory, designed so that each of the probes 00:03:18.810 --> 00:03:21.960 align:middle line:84% has similar characteristics in terms of melting temperature, 00:03:21.960 --> 00:03:24.070 align:middle line:84% and some of these other characteristics. 00:03:24.070 --> 00:03:27.790 align:middle line:84% So in theory you can compare one spot to another spot. 00:03:27.790 --> 00:03:29.580 align:middle line:84% And because of that, these are made-- 00:03:29.580 --> 00:03:32.110 align:middle line:84% these are hybridized using a single sample. 00:03:32.110 --> 00:03:34.680 align:middle line:84% So you put one sample on one array here, 00:03:34.680 --> 00:03:38.230 align:middle line:84% and if you want to do a relative difference you use two arrays. 00:03:38.230 --> 00:03:42.000 align:middle line:84% Now again in theory this is an absolute expression amount. 00:03:42.000 --> 00:03:45.630 align:middle line:84% If people want to get to the most absolute measurement, 00:03:45.630 --> 00:03:49.990 align:middle line:84% like how many strands of RNA are present, 00:03:49.990 --> 00:03:52.050 align:middle line:84% then there's another technology called SAGE, 00:03:52.050 --> 00:03:58.050 align:middle line:84% and that's how I just wanted to introduce that discussion here. 00:03:58.050 --> 00:04:01.330 align:middle line:84% SAGE stands for serial analysis of gene expression. 00:04:01.330 --> 00:04:03.970 align:middle line:84% I think the original paper on this was in Science. 00:04:03.970 --> 00:04:07.300 align:middle line:84% It was a one or two-page paper back in 1995. 00:04:07.300 --> 00:04:12.080 align:middle line:84% by-- it was Victor Velculescu at Hopkins. 00:04:12.080 --> 00:04:13.672 align:middle line:84% And so this actually was developed 00:04:13.672 --> 00:04:15.130 align:middle line:84% around the same time as microarrays 00:04:15.130 --> 00:04:17.410 align:middle line:84% but actually gained a lot of popularity 00:04:17.410 --> 00:04:19.600 align:middle line:84% before microarrays became so commonly available. 00:04:19.600 --> 00:04:22.690 align:middle line:84% But essentially the way SAGE works is that it gives you 00:04:22.690 --> 00:04:25.540 align:middle line:84% an absolute expression level for a particular gene, 00:04:25.540 --> 00:04:28.990 align:middle line:84% but it measures gene expression by sequencing. 00:04:28.990 --> 00:04:31.570 align:middle line:84% And automated sequencers were just coming around at the time, 00:04:31.570 --> 00:04:33.730 align:middle line:84% and people realized that you can use this technique 00:04:33.730 --> 00:04:36.880 align:middle line:84% to use them to measure RNA expression levels. 00:04:36.880 --> 00:04:41.170 align:middle line:84% So the way this works is this, is that in theory every RNA-- 00:04:41.170 --> 00:04:43.960 align:middle line:84% or let's say every sequence, every gene-- 00:04:43.960 --> 00:04:47.410 align:middle line:84% has what they call, is a SAGE tag here. 00:04:47.410 --> 00:04:51.760 align:middle line:84% And all the SAGE tag is is, let's say, 10 nucleotides 00:04:51.760 --> 00:04:55.660 align:middle line:84% that hopefully uniquely identify that gene. 00:04:55.660 --> 00:04:59.410 align:middle line:84% But the catch is it has to be after a CATG, 00:04:59.410 --> 00:05:01.670 align:middle line:90% and we'll see why in a moment. 00:05:01.670 --> 00:05:05.110 align:middle line:84% So you look at a gene, you look for a CATG, 00:05:05.110 --> 00:05:08.050 align:middle line:84% and the most downstream 10 nucleotides 00:05:08.050 --> 00:05:10.030 align:middle line:84% are going to be the tag for that gene. 00:05:10.030 --> 00:05:12.800 align:middle line:84% Hopefully that's enough to make it unique, 00:05:12.800 --> 00:05:16.040 align:middle line:84% but that's not always going to be the case. 00:05:16.040 --> 00:05:17.290 align:middle line:90% So the way this works is this. 00:05:17.290 --> 00:05:19.570 align:middle line:84% If you start with the strands of RNA 00:05:19.570 --> 00:05:22.150 align:middle line:84% that you're going to measure, the first thing you want to do 00:05:22.150 --> 00:05:24.400 align:middle line:90% is to make a strand of DNA. 00:05:24.400 --> 00:05:25.833 align:middle line:90% RNA is very fragile. 00:05:25.833 --> 00:05:28.000 align:middle line:84% You want to convert it to DNA as quickly as you can. 00:05:28.000 --> 00:05:30.950 align:middle line:84% Otherwise it's going to start to degrade. 00:05:30.950 --> 00:05:33.387 align:middle line:84% So the techniques for taking a strand of RNA 00:05:33.387 --> 00:05:35.470 align:middle line:84% and making a strand of DNA are commonly available. 00:05:35.470 --> 00:05:37.180 align:middle line:90% It's a reverse transcriptase. 00:05:37.180 --> 00:05:40.090 align:middle line:84% You can buy this off the shelf, or you can buy it from a store. 00:05:40.090 --> 00:05:42.880 align:middle line:84% And essentially it's going to start by going at one end. 00:05:42.880 --> 00:05:44.710 align:middle line:84% You might give it a primer to get started, 00:05:44.710 --> 00:05:46.690 align:middle line:84% and then it's just going to go all the way until it falls off 00:05:46.690 --> 00:05:47.320 align:middle line:90% the end. 00:05:47.320 --> 00:05:50.660 align:middle line:84% So now you've made a cDNA copy of this. 00:05:50.660 --> 00:05:53.220 align:middle line:84% And what you do now is, you can do that for the other strand 00:05:53.220 --> 00:05:53.720 align:middle line:90% as well. 00:05:53.720 --> 00:05:56.350 align:middle line:84% So now you've gotten rid of the RNA, and you have two, 00:05:56.350 --> 00:05:59.680 align:middle line:84% you have a double strand of DNA representing the RNA 00:05:59.680 --> 00:06:02.680 align:middle line:84% that you had in your original sample. 00:06:02.680 --> 00:06:06.340 align:middle line:84% Now it gets a little tricky, because now we're essentially 00:06:06.340 --> 00:06:08.350 align:middle line:84% going to use a lot of molecular tools 00:06:08.350 --> 00:06:12.340 align:middle line:84% that bacteria have to actually do this kind of fancy cutting 00:06:12.340 --> 00:06:14.565 align:middle line:84% and pasting so that we can sequence this 00:06:14.565 --> 00:06:15.940 align:middle line:84% and figure out what the genes are 00:06:15.940 --> 00:06:17.783 align:middle line:84% and what the gene expression level is. 00:06:17.783 --> 00:06:19.450 align:middle line:84% There's a particular restriction enzyme. 00:06:19.450 --> 00:06:22.953 align:middle line:84% So a restriction enzyme is actually a component. 00:06:22.953 --> 00:06:24.370 align:middle line:84% It's a protein certainly, but it's 00:06:24.370 --> 00:06:26.410 align:middle line:84% an enzyme that's present in bacteria 00:06:26.410 --> 00:06:30.422 align:middle line:84% that goes around cutting DNA at particular sequences. 00:06:30.422 --> 00:06:32.380 align:middle line:84% There's a lot of restriction enzymes out there. 00:06:32.380 --> 00:06:34.960 align:middle line:84% If you get the current catalog or the current poster, 00:06:34.960 --> 00:06:37.570 align:middle line:84% there might be on the order of, let's say, about 300 00:06:37.570 --> 00:06:39.280 align:middle line:90% of these restriction enzymes. 00:06:39.280 --> 00:06:41.230 align:middle line:84% And there are many ways to define them, 00:06:41.230 --> 00:06:43.450 align:middle line:84% but one way people define restriction enzymes is 00:06:43.450 --> 00:06:45.670 align:middle line:90% based on the sequence it cuts. 00:06:45.670 --> 00:06:48.010 align:middle line:84% And so there's some that cut DNA, 00:06:48.010 --> 00:06:50.110 align:middle line:84% giving absolutely flat ends, and there's 00:06:50.110 --> 00:06:52.780 align:middle line:84% some that actually cut leaving sticky ends. 00:06:52.780 --> 00:06:54.880 align:middle line:84% And so here, this particular restriction enzyme 00:06:54.880 --> 00:07:00.770 align:middle line:84% called NLA3 takes any strand of double-stranded DNA 00:07:00.770 --> 00:07:05.710 align:middle line:84% and cuts it at the CATG, but it leaves the CATG as an overhang. 00:07:05.710 --> 00:07:07.938 align:middle line:90% It doesn't cut it flat. 00:07:07.938 --> 00:07:09.730 align:middle line:84% So these guys saw this, and they said, wow, 00:07:09.730 --> 00:07:12.230 align:middle line:84% we could do some interesting things, a little bit of cutting 00:07:12.230 --> 00:07:14.510 align:middle line:84% and pasting, and actually take advantage of this. 00:07:14.510 --> 00:07:17.150 align:middle line:84% So to be clear, here's the sequence. 00:07:17.150 --> 00:07:19.210 align:middle line:84% Here's the CATG and the unique tag. 00:07:19.210 --> 00:07:21.790 align:middle line:84% And basically we're going to-- even though the strand might 00:07:21.790 --> 00:07:24.025 align:middle line:84% be longer than this, it's going to get cut like this. 00:07:24.025 --> 00:07:26.860 align:middle line:90% 00:07:26.860 --> 00:07:29.480 align:middle line:84% Now to use some more fancy parts to this, 00:07:29.480 --> 00:07:33.400 align:middle line:84% remember how I told you most messenger RNA has a strand of A 00:07:33.400 --> 00:07:35.980 align:middle line:84% at the end, because as it's being copied off the DNA 00:07:35.980 --> 00:07:39.210 align:middle line:84% it starts to stutter in some ways and just leaves a poly(A). 00:07:39.210 --> 00:07:42.220 align:middle line:84% And that you can bind to it, or you can actually 00:07:42.220 --> 00:07:45.448 align:middle line:90% find it by using a poly(T). 00:07:45.448 --> 00:07:46.990 align:middle line:84% Using their technique what you can do 00:07:46.990 --> 00:07:48.448 align:middle line:84% is instead of just starting with T, 00:07:48.448 --> 00:07:52.210 align:middle line:84% start it with Ts and something sticky on the end. 00:07:52.210 --> 00:07:54.160 align:middle line:84% In other words, it's labeled in such a way 00:07:54.160 --> 00:07:57.850 align:middle line:84% that we can then fish these out after we apply 00:07:57.850 --> 00:08:00.220 align:middle line:90% a magnet to it, for example. 00:08:00.220 --> 00:08:05.062 align:middle line:84% Or some people use streptavidin biotinylation as a common way 00:08:05.062 --> 00:08:06.520 align:middle line:84% to actually pull, fish these things 00:08:06.520 --> 00:08:07.937 align:middle line:84% out when you run through a column, 00:08:07.937 --> 00:08:12.410 align:middle line:84% for example, with particular beads, however you do it. 00:08:12.410 --> 00:08:15.720 align:middle line:84% So now we have a whole bunch of the double-stranded DNA, 00:08:15.720 --> 00:08:17.643 align:middle line:84% except we have this little overhang here. 00:08:17.643 --> 00:08:19.310 align:middle line:84% We've got all the As and Ts on that end. 00:08:19.310 --> 00:08:21.820 align:middle line:84% We fished them out of this big cellular pool, 00:08:21.820 --> 00:08:24.820 align:middle line:90% this cellular mess. 00:08:24.820 --> 00:08:26.160 align:middle line:90% Now what we're going to do-- 00:08:26.160 --> 00:08:27.535 align:middle line:84% it starts to get a little tricky. 00:08:27.535 --> 00:08:29.360 align:middle line:90% But now we have this tag here. 00:08:29.360 --> 00:08:30.580 align:middle line:90% Keep your eye on the red. 00:08:30.580 --> 00:08:32.413 align:middle line:84% That's all I can say here in this technique. 00:08:32.413 --> 00:08:34.330 align:middle line:84% Because what you're going to do is, in the end 00:08:34.330 --> 00:08:36.220 align:middle line:84% you're going to see how we're going 00:08:36.220 --> 00:08:39.370 align:middle line:84% to be able to use a sequencer to tell these RNAs how 00:08:39.370 --> 00:08:41.390 align:middle line:90% much of each RNA is present. 00:08:41.390 --> 00:08:43.120 align:middle line:84% So we've got one RNA going this way, 00:08:43.120 --> 00:08:45.430 align:middle line:84% and we actually got this light blue for another RNA 00:08:45.430 --> 00:08:46.160 align:middle line:90% going this way. 00:08:46.160 --> 00:08:47.140 align:middle line:84% And you're going to see how they're going to end up 00:08:47.140 --> 00:08:49.100 align:middle line:90% being next to each other. 00:08:49.100 --> 00:08:50.680 align:middle line:84% So what we have to do then is add 00:08:50.680 --> 00:08:53.380 align:middle line:84% an adapter that has another sticky end here, 00:08:53.380 --> 00:08:56.290 align:middle line:90% and add it to this side here. 00:08:56.290 --> 00:09:00.180 align:middle line:84% And we're going to use another restriction enzyme that's 00:09:00.180 --> 00:09:02.550 align:middle line:84% going to cut at the CATG, but it's 00:09:02.550 --> 00:09:03.992 align:middle line:90% going to cut after the tag. 00:09:03.992 --> 00:09:05.700 align:middle line:84% There's one particular restriction enzyme 00:09:05.700 --> 00:09:10.590 align:middle line:84% called BsmF1 that looks for CATG but doesn't cut at the CATG. 00:09:10.590 --> 00:09:13.590 align:middle line:84% It cuts 14 base pairs downstream from there. 00:09:13.590 --> 00:09:17.550 align:middle line:84% That's why our tags have to be 10 nucleotides long. 00:09:17.550 --> 00:09:21.990 align:middle line:84% A lot of fancy biochemistry here to get this to work. 00:09:21.990 --> 00:09:23.910 align:middle line:84% So you got a tag here, and you got tag here, 00:09:23.910 --> 00:09:25.285 align:middle line:84% and you got a little adapter here 00:09:25.285 --> 00:09:28.560 align:middle line:90% that we've added on this side. 00:09:28.560 --> 00:09:34.560 align:middle line:84% And then finally, what we're going to do, we've cut it here, 00:09:34.560 --> 00:09:36.240 align:middle line:90% and we've cut it here. 00:09:36.240 --> 00:09:39.360 align:middle line:84% What we can do is add something called a ligase that basically 00:09:39.360 --> 00:09:43.260 align:middle line:84% takes all blunt ends and just tries to make pairs together. 00:09:43.260 --> 00:09:46.470 align:middle line:84% It tries to bring any small strands of DNA with blunt ends 00:09:46.470 --> 00:09:48.168 align:middle line:84% and tries to attach them together. 00:09:48.168 --> 00:09:49.710 align:middle line:84% So we have a blunt end that was here, 00:09:49.710 --> 00:09:51.010 align:middle line:90% and a blunt end that was here. 00:09:51.010 --> 00:09:53.580 align:middle line:84% Here's the first tag, and here's the second tag. 00:09:53.580 --> 00:09:55.650 align:middle line:84% And we're going to have a whole bunch of these. 00:09:55.650 --> 00:09:57.990 align:middle line:84% And what we can do is just run it through a sequencer, 00:09:57.990 --> 00:09:58.740 align:middle line:90% basically. 00:09:58.740 --> 00:10:01.170 align:middle line:84% We're going to do some more fancy footwork to get these 00:10:01.170 --> 00:10:03.700 align:middle line:90% to be as compact as possible. 00:10:03.700 --> 00:10:06.000 align:middle line:84% And in the end we have one stretch 00:10:06.000 --> 00:10:09.420 align:middle line:84% of DNA with just one tag after another, 00:10:09.420 --> 00:10:12.180 align:middle line:84% with some CATGs in the middle as spacers 00:10:12.180 --> 00:10:15.390 align:middle line:84% so we can make sure we're staying synchronized. 00:10:15.390 --> 00:10:18.030 align:middle line:84% Now what you do is run this through an automated sequencer, 00:10:18.030 --> 00:10:20.790 align:middle line:84% and then you can just get-- you can read off the sequence just 00:10:20.790 --> 00:10:23.310 align:middle line:84% like this, one then the other, then the one then 00:10:23.310 --> 00:10:25.860 align:middle line:84% the other, and then search databases 00:10:25.860 --> 00:10:29.790 align:middle line:84% to see what gene do we think this is. 00:10:29.790 --> 00:10:32.190 align:middle line:84% Now the reason why this is an absolute gene expression 00:10:32.190 --> 00:10:34.500 align:middle line:84% measurement, for sure, is that if this gene is 00:10:34.500 --> 00:10:38.190 align:middle line:84% over-represented, it's two times more than this one, 00:10:38.190 --> 00:10:41.040 align:middle line:84% you're going to see two times more tags than this one, 00:10:41.040 --> 00:10:43.200 align:middle line:90% in theory. 00:10:43.200 --> 00:10:45.815 align:middle line:84% There are efficiencies to each one of these steps 00:10:45.815 --> 00:10:46.690 align:middle line:90% that we talked about. 00:10:46.690 --> 00:10:49.290 align:middle line:84% And this is not as simple as a microarray. 00:10:49.290 --> 00:10:50.965 align:middle line:84% Literally you can take the sample 00:10:50.965 --> 00:10:53.340 align:middle line:84% if you're a physician at Brigham Women's Hospital, or any 00:10:53.340 --> 00:10:55.548 align:middle line:84% of these hospitals, and essentially take it to a core 00:10:55.548 --> 00:10:56.760 align:middle line:90% and get an microarray done. 00:10:56.760 --> 00:11:00.060 align:middle line:84% SAGE is a lot more involved, but in theory 00:11:00.060 --> 00:11:02.880 align:middle line:84% this is a great way to pick up low expression-- 00:11:02.880 --> 00:11:05.748 align:middle line:84% genes with very low expression levels. 00:11:05.748 --> 00:11:08.040 align:middle line:84% So that's why it was worth at least talking about SAGE. 00:11:08.040 --> 00:11:10.373 align:middle line:84% Because if you're looking at absolute expression levels, 00:11:10.373 --> 00:11:11.760 align:middle line:90% this is where it's going. 00:11:11.760 --> 00:11:14.220 align:middle line:84% For now this is a back burner kind of technology. 00:11:14.220 --> 00:11:15.250 align:middle line:90% People still use it. 00:11:15.250 --> 00:11:17.490 align:middle line:84% There are people who are religiously devoted to it, 00:11:17.490 --> 00:11:20.250 align:middle line:84% but most people end up using microarrays today. 00:11:20.250 --> 00:11:20.750 align:middle line:90% Yeah? 00:11:20.750 --> 00:11:22.200 align:middle line:84% AUDIENCE: Can you go back and explain just a little bit 00:11:22.200 --> 00:11:23.850 align:middle line:90% how the tag got there that way? 00:11:23.850 --> 00:11:27.340 align:middle line:84% ATUL BUTTE: The tag just happens to be a sequence. 00:11:27.340 --> 00:11:30.300 align:middle line:84% So what we're calling a tag here is nothing we put there. 00:11:30.300 --> 00:11:31.080 align:middle line:90% AUDIENCE: OK. 00:11:31.080 --> 00:11:34.950 align:middle line:84% ATUL BUTTE: It's something that is next to a CATG that's 00:11:34.950 --> 00:11:37.590 align:middle line:90% on this end of the gene. 00:11:37.590 --> 00:11:40.080 align:middle line:84% If there's nothing there like that, then we're 00:11:40.080 --> 00:11:43.110 align:middle line:84% not going to find it, so that's another catch. 00:11:43.110 --> 00:11:44.820 align:middle line:84% If the first one is all the way up here, 00:11:44.820 --> 00:11:46.350 align:middle line:90% we might not even get it. 00:11:46.350 --> 00:11:49.170 align:middle line:84% So they're saying the average, if you do the math, 00:11:49.170 --> 00:11:51.930 align:middle line:84% there's at least one of these within 256 base pairs, 00:11:51.930 --> 00:11:54.060 align:middle line:84% because it's like 4 to the fourth. 00:11:54.060 --> 00:11:56.850 align:middle line:84% So there should be one there, but there might be degeneracy, 00:11:56.850 --> 00:11:59.200 align:middle line:90% or it might not be there. 00:11:59.200 --> 00:12:00.450 align:middle line:90% So that's the whole technique. 00:12:00.450 --> 00:12:01.260 align:middle line:90% It's 10. 00:12:01.260 --> 00:12:03.510 align:middle line:84% It's CATG only because of the peculiarities 00:12:03.510 --> 00:12:06.210 align:middle line:84% of the enzymes being used to actually cut these things apart 00:12:06.210 --> 00:12:08.820 align:middle line:90% and then just look at the tags. 00:12:08.820 --> 00:12:10.545 align:middle line:90% Any questions about that? 00:12:10.545 --> 00:12:12.420 align:middle line:84% So that's basically-- that was the last thing 00:12:12.420 --> 00:12:15.330 align:middle line:84% I want to talk about in this section of gene measurement 00:12:15.330 --> 00:12:18.040 align:middle line:84% technologies, or gene measurement techniques. 00:12:18.040 --> 00:12:22.200 align:middle line:84% So now Zack was suggesting that I talk about some of the stuff 00:12:22.200 --> 00:12:23.700 align:middle line:84% that we've been doing with diabetes, 00:12:23.700 --> 00:12:26.760 align:middle line:84% only because this is supposed to be genomic medicine. 00:12:26.760 --> 00:12:28.500 align:middle line:84% We can give you a real world example 00:12:28.500 --> 00:12:34.120 align:middle line:84% of how we've used microarrays to help us with diabetes. 00:12:34.120 --> 00:12:36.133 align:middle line:84% So I'm a pediatric endocrinologist, 00:12:36.133 --> 00:12:38.550 align:middle line:84% and I've spent most of my time over at the Joslin Diabetes 00:12:38.550 --> 00:12:40.065 align:middle line:90% Center studying type 2 diabetes. 00:12:40.065 --> 00:12:42.960 align:middle line:84% It currently affects about 15 million people in the United 00:12:42.960 --> 00:12:44.790 align:middle line:84% States, and in fact, the CDC just 00:12:44.790 --> 00:12:47.880 align:middle line:84% estimated a couple of months ago that a child born in the year 00:12:47.880 --> 00:12:52.450 align:middle line:84% 2000 now has a 1 in 3 risk of getting type 2 diabetes. 00:12:52.450 --> 00:12:56.550 align:middle line:84% So it's 33% likelihood now of a child born just two years ago 00:12:56.550 --> 00:12:57.750 align:middle line:90% to get type 2 diabetes. 00:12:57.750 --> 00:12:58.290 align:middle line:90% Why? 00:12:58.290 --> 00:13:00.570 align:middle line:90% Because of the obesity problem. 00:13:00.570 --> 00:13:05.130 align:middle line:84% So we all know kids have less exercise today in school, 00:13:05.130 --> 00:13:06.810 align:middle line:84% because they have to pass these tests. 00:13:06.810 --> 00:13:08.807 align:middle line:84% And now schools have to give more courses 00:13:08.807 --> 00:13:10.890 align:middle line:84% so that they pass the MCASes and things like that. 00:13:10.890 --> 00:13:11.893 align:middle line:90% There's more TV. 00:13:11.893 --> 00:13:12.810 align:middle line:90% There's more channels. 00:13:12.810 --> 00:13:13.660 align:middle line:90% There's internet. 00:13:13.660 --> 00:13:15.060 align:middle line:84% There's all these things that cause kids 00:13:15.060 --> 00:13:16.310 align:middle line:90% to basically sit in one place. 00:13:16.310 --> 00:13:17.730 align:middle line:90% They don't get much exercise. 00:13:17.730 --> 00:13:19.147 align:middle line:84% I think there was something on CNN 00:13:19.147 --> 00:13:21.142 align:middle line:84% last week about how even toddlers one or two 00:13:21.142 --> 00:13:23.100 align:middle line:84% years old are just not getting as much exercise 00:13:23.100 --> 00:13:24.190 align:middle line:90% as they used to. 00:13:24.190 --> 00:13:25.440 align:middle line:90% So this isn't even school age. 00:13:25.440 --> 00:13:28.650 align:middle line:84% This is starting at basically age one or two. 00:13:28.650 --> 00:13:31.560 align:middle line:84% So the question we had here was this. 00:13:31.560 --> 00:13:33.450 align:middle line:90% We know how to define-- 00:13:33.450 --> 00:13:36.960 align:middle line:84% the diagnosis of diabetes is actually sort of arbitrary. 00:13:36.960 --> 00:13:39.540 align:middle line:84% In fact, the definition of diabetes keeps changing, 00:13:39.540 --> 00:13:40.560 align:middle line:90% it turns out. 00:13:40.560 --> 00:13:43.260 align:middle line:84% But for example, at one point the definition of diabetes was 00:13:43.260 --> 00:13:46.320 align:middle line:84% having a fasting blood sugar-- so morning blood sugar-- 00:13:46.320 --> 00:13:49.710 align:middle line:84% level of 127, I think it was, milligrams per deciliter 00:13:49.710 --> 00:13:52.290 align:middle line:84% qualified you for having diabetes. 00:13:52.290 --> 00:13:54.060 align:middle line:84% But what we know is that it's not just 00:13:54.060 --> 00:13:55.593 align:middle line:90% an on and off type of thing. 00:13:55.593 --> 00:13:57.260 align:middle line:84% All of a sudden you don't have diabetes, 00:13:57.260 --> 00:13:59.030 align:middle line:90% and now all of a sudden you do. 00:13:59.030 --> 00:14:03.230 align:middle line:84% Because diabetes is a very heritable trait. 00:14:03.230 --> 00:14:04.760 align:middle line:84% It's controversial, but some people 00:14:04.760 --> 00:14:06.593 align:middle line:84% say, for example, it's a very genetic trait. 00:14:06.593 --> 00:14:07.970 align:middle line:90% Let's put it that way. 00:14:07.970 --> 00:14:11.210 align:middle line:84% Some studies that show that in twin-twin studies if you have 00:14:11.210 --> 00:14:12.990 align:middle line:84% one twin that has type 2 diabetes, 00:14:12.990 --> 00:14:15.110 align:middle line:84% you have a 98% likelihood that the other twin is 00:14:15.110 --> 00:14:17.780 align:middle line:84% going to get it, if they don't already have it. 00:14:17.780 --> 00:14:20.570 align:middle line:84% So it makes you think that it's incredibly genetic, 00:14:20.570 --> 00:14:22.470 align:middle line:84% but the problem is the incidence is so high. 00:14:22.470 --> 00:14:24.080 align:middle line:90% This is a common disease. 00:14:24.080 --> 00:14:27.440 align:middle line:84% It's not very easy at all to find a particular mutation that 00:14:27.440 --> 00:14:28.010 align:middle line:90% causes it. 00:14:28.010 --> 00:14:30.560 align:middle line:84% We haven't been that lucky so far. 00:14:30.560 --> 00:14:33.830 align:middle line:84% We have found-- the field has found a number of mutations 00:14:33.830 --> 00:14:36.020 align:middle line:84% that do lead to diabetes, but they 00:14:36.020 --> 00:14:40.370 align:middle line:84% might explain the sum total of maybe 0.01% of the people 00:14:40.370 --> 00:14:41.600 align:middle line:90% who have diabetes. 00:14:41.600 --> 00:14:42.890 align:middle line:90% Those are very rare forms. 00:14:42.890 --> 00:14:46.490 align:middle line:84% The common form, which seems like it's incredibly genetic, 00:14:46.490 --> 00:14:48.698 align:middle line:90% it's still a big mystery. 00:14:48.698 --> 00:14:50.990 align:middle line:84% The point of what we were going to use genomic medicine 00:14:50.990 --> 00:14:54.320 align:middle line:84% for here was trying to define the patterns of how 00:14:54.320 --> 00:14:57.470 align:middle line:84% people might go from non-diabetes to diabetes. 00:14:57.470 --> 00:14:59.548 align:middle line:90% There's a middle group in there. 00:14:59.548 --> 00:15:01.340 align:middle line:84% In fact, the ADA, the Diabetes Association, 00:15:01.340 --> 00:15:03.440 align:middle line:84% now calls this pre-diabetes, people 00:15:03.440 --> 00:15:06.830 align:middle line:84% who don't have a normal handling of sugar 00:15:06.830 --> 00:15:11.630 align:middle line:84% but don't strictly meet the criteria for diabetes today. 00:15:11.630 --> 00:15:14.240 align:middle line:84% The question we had here, using microarrays, 00:15:14.240 --> 00:15:16.520 align:middle line:84% can we define patterns of gene expression 00:15:16.520 --> 00:15:19.340 align:middle line:84% in human subjects with diabetes, specifically in their muscle 00:15:19.340 --> 00:15:22.580 align:middle line:84% samples, to identify those that are at high 00:15:22.580 --> 00:15:26.420 align:middle line:84% risk not just define what genes are different between diabetes 00:15:26.420 --> 00:15:29.810 align:middle line:84% and non-diabetes but also what genes were already different 00:15:29.810 --> 00:15:32.570 align:middle line:84% in those that were at high risk of getting diabetes. 00:15:32.570 --> 00:15:36.405 align:middle line:84% Specifically these are offspring of type 2 diabetics. 00:15:36.405 --> 00:15:38.030 align:middle line:84% So in this particular study that we did 00:15:38.030 --> 00:15:39.710 align:middle line:84% was in Mexican-American subjects. 00:15:39.710 --> 00:15:43.698 align:middle line:84% We have diabetes and family history of diabetes. 00:15:43.698 --> 00:15:45.740 align:middle line:84% So we basically have three groups, family history 00:15:45.740 --> 00:15:48.142 align:middle line:90% negative, diabetes-- 00:15:48.142 --> 00:15:49.850 align:middle line:84% so obviously these are control-- and then 00:15:49.850 --> 00:15:52.010 align:middle line:84% we have family history positives. 00:15:52.010 --> 00:15:53.780 align:middle line:90% So we got three groups here. 00:15:53.780 --> 00:15:56.938 align:middle line:84% And you can look at the subjects that we 00:15:56.938 --> 00:15:58.230 align:middle line:90% used for this particular study. 00:15:58.230 --> 00:16:00.375 align:middle line:84% We had one that we use for the arrays, 00:16:00.375 --> 00:16:02.000 align:middle line:84% and we had another that we actually did 00:16:02.000 --> 00:16:04.970 align:middle line:84% validation work on, some of the hypotheses that we came up 00:16:04.970 --> 00:16:05.760 align:middle line:90% with. 00:16:05.760 --> 00:16:07.260 align:middle line:84% And so most of these characteristics 00:16:07.260 --> 00:16:09.468 align:middle line:84% are not different, but they are different in a number 00:16:09.468 --> 00:16:10.260 align:middle line:90% of important ways. 00:16:10.260 --> 00:16:14.960 align:middle line:84% So fasting glucose is 99, 92, but 200 in the diabetics 00:16:14.960 --> 00:16:17.270 align:middle line:84% obviously, because they have diabetes. 00:16:17.270 --> 00:16:20.780 align:middle line:84% The two-hour glucose was higher in the diabetics, 00:16:20.780 --> 00:16:23.930 align:middle line:84% but the hemoglobin A1C was already high in the family 00:16:23.930 --> 00:16:24.900 align:middle line:90% history positives. 00:16:24.900 --> 00:16:27.272 align:middle line:90% What is the hemoglobin A1C? 00:16:27.272 --> 00:16:28.730 align:middle line:84% Well that's one of our measurements 00:16:28.730 --> 00:16:31.760 align:middle line:84% for how high the blood sugar levels are 00:16:31.760 --> 00:16:34.410 align:middle line:90% over a span of 120 days. 00:16:34.410 --> 00:16:37.010 align:middle line:84% So in fact most physicians now who 00:16:37.010 --> 00:16:38.630 align:middle line:84% take care of patients with diabetes, 00:16:38.630 --> 00:16:41.523 align:middle line:84% we measure this hemoglobin A1C at least four times a year 00:16:41.523 --> 00:16:43.940 align:middle line:84% so that we can get a gauge as to what the blood sugars are 00:16:43.940 --> 00:16:46.850 align:middle line:84% like over the past 120 days, because that's the lifespan 00:16:46.850 --> 00:16:48.830 align:middle line:90% of a erythrocyte or a red cell. 00:16:48.830 --> 00:16:50.960 align:middle line:84% And as the glucose levels are higher in the blood, 00:16:50.960 --> 00:16:53.700 align:middle line:84% they actually glycosylate the hemoglobin in the red cell, 00:16:53.700 --> 00:16:56.270 align:middle line:90% so it's a byproduct of this. 00:16:56.270 --> 00:16:59.270 align:middle line:84% But it's already high in the family history positive. 00:16:59.270 --> 00:17:03.950 align:middle line:84% They're not exactly like the family history negatives here. 00:17:03.950 --> 00:17:06.440 align:middle line:90% Fasting insulin already high. 00:17:06.440 --> 00:17:08.023 align:middle line:84% So the problem with type 2 diabetes-- 00:17:08.023 --> 00:17:10.190 align:middle line:84% just to make sure everyone's on the same page here-- 00:17:10.190 --> 00:17:13.010 align:middle line:84% it's not a problem with making the insulin from the pancreas. 00:17:13.010 --> 00:17:14.780 align:middle line:84% The problem is that insulin is there, 00:17:14.780 --> 00:17:17.780 align:middle line:84% it just doesn't act like it's supposed to in the target 00:17:17.780 --> 00:17:19.369 align:middle line:90% tissues. 00:17:19.369 --> 00:17:21.170 align:middle line:84% Now to use my analogy from last week, 00:17:21.170 --> 00:17:25.550 align:middle line:84% I just had this great food from the Chinese truck again. 00:17:25.550 --> 00:17:28.920 align:middle line:84% My glucose levels are going up, my insulin levels are going up. 00:17:28.920 --> 00:17:31.748 align:middle line:84% And so insulin is now telling a number of my tissues 00:17:31.748 --> 00:17:32.790 align:middle line:90% to do a number of things. 00:17:32.790 --> 00:17:35.870 align:middle line:84% It's telling my liver to stop making sugar. 00:17:35.870 --> 00:17:37.310 align:middle line:90% I just ate a whole huge meal. 00:17:37.310 --> 00:17:39.590 align:middle line:84% I don't need the liver to start making more sugar, 00:17:39.590 --> 00:17:41.143 align:middle line:84% because I just ate a whole bunch. 00:17:41.143 --> 00:17:42.560 align:middle line:84% But more importantly, it's telling 00:17:42.560 --> 00:17:47.480 align:middle line:84% that my muscle and my fat cells to start taking in the sugar. 00:17:47.480 --> 00:17:49.130 align:middle line:84% And what happens in type 2 diabetes 00:17:49.130 --> 00:17:51.740 align:middle line:84% is that that signal from the insulin to what the tissues are 00:17:51.740 --> 00:17:54.770 align:middle line:84% supposed to do is muted for some reason. 00:17:54.770 --> 00:17:56.930 align:middle line:90% We just don't know exactly why. 00:17:56.930 --> 00:17:57.950 align:middle line:90% But that's the problem. 00:17:57.950 --> 00:17:59.160 align:middle line:90% The insulin is there. 00:17:59.160 --> 00:18:02.090 align:middle line:84% In fact, now these guys have a higher insulin level, 00:18:02.090 --> 00:18:05.900 align:middle line:84% because already their bodies are realizing 00:18:05.900 --> 00:18:09.440 align:middle line:84% the same amount of insulin isn't doing what it's supposed to. 00:18:09.440 --> 00:18:13.100 align:middle line:84% So the pancreas sees that the sugar levels are not going down 00:18:13.100 --> 00:18:15.530 align:middle line:84% after the meals, for example, and it's saying, 00:18:15.530 --> 00:18:18.350 align:middle line:84% I need to ramp up the amount of insulin I'm making. 00:18:18.350 --> 00:18:20.870 align:middle line:84% But the primary problem is felt to be in the target 00:18:20.870 --> 00:18:24.530 align:middle line:90% tissues, insulin resistance. 00:18:24.530 --> 00:18:26.030 align:middle line:84% So the groups are already different, 00:18:26.030 --> 00:18:29.820 align:middle line:84% and that's going to actually make this analysis complicated. 00:18:29.820 --> 00:18:33.627 align:middle line:84% So we have five diabetics there, four without diabetes 00:18:33.627 --> 00:18:35.210 align:middle line:84% but with a family history of diabetes, 00:18:35.210 --> 00:18:37.500 align:middle line:90% and six controls here. 00:18:37.500 --> 00:18:39.950 align:middle line:84% And when you do this, you can compare each group 00:18:39.950 --> 00:18:40.950 align:middle line:90% to each other group. 00:18:40.950 --> 00:18:44.300 align:middle line:84% So we can take-- we took muscle samples from these individuals. 00:18:44.300 --> 00:18:46.310 align:middle line:84% If they were on medicines for the diabetics, 00:18:46.310 --> 00:18:48.393 align:middle line:84% they were actually taken off for a number of days. 00:18:48.393 --> 00:18:50.690 align:middle line:84% They had no strenuous exercise beforehand. 00:18:50.690 --> 00:18:53.990 align:middle line:84% And it's all informed consent, and you get the muscle samples. 00:18:53.990 --> 00:18:55.555 align:middle line:90% You extract out the RNA. 00:18:55.555 --> 00:18:56.930 align:middle line:84% We put them on these microarrays. 00:18:56.930 --> 00:18:59.180 align:middle line:84% The microarrays when we did these experiments 00:18:59.180 --> 00:19:01.010 align:middle line:90% measured about 7,000 genes. 00:19:01.010 --> 00:19:02.960 align:middle line:84% Today when you do this for the same price 00:19:02.960 --> 00:19:05.990 align:middle line:90% you get 44,000 genes. 00:19:05.990 --> 00:19:09.147 align:middle line:84% And so we can use a number of different analytic techniques 00:19:09.147 --> 00:19:10.730 align:middle line:84% that you're going to be learning about 00:19:10.730 --> 00:19:13.788 align:middle line:84% over the rest of the semester-- exactly how we make 00:19:13.788 --> 00:19:15.830 align:middle line:84% the distinction of what genes might be different, 00:19:15.830 --> 00:19:17.558 align:middle line:90% what genes are not changing. 00:19:17.558 --> 00:19:19.850 align:middle line:84% But if you compare family history negative to diabetes, 00:19:19.850 --> 00:19:22.010 align:middle line:84% we find 187 genes that are different. 00:19:22.010 --> 00:19:24.950 align:middle line:84% Even if you compare family history positive to diabetes, 00:19:24.950 --> 00:19:28.030 align:middle line:84% there's 166 genes that are different. 00:19:28.030 --> 00:19:30.000 align:middle line:84% So fewer, and there's some overlap. 00:19:30.000 --> 00:19:32.160 align:middle line:90% So the 55 genes between-- 00:19:32.160 --> 00:19:36.970 align:middle line:84% there are 55 genes that were in this list and in this list. 00:19:36.970 --> 00:19:40.110 align:middle line:84% Most of them coded for all sorts of interesting proteins. 00:19:40.110 --> 00:19:41.640 align:middle line:84% For example, mitochondrial proteins 00:19:41.640 --> 00:19:45.338 align:middle line:84% that are involved in actually energy metabolism, or ATP 00:19:45.338 --> 00:19:47.130 align:middle line:84% synthesis, oxidative phosphorylation, which 00:19:47.130 --> 00:19:50.320 align:middle line:90% we'll talk about in a moment. 00:19:50.320 --> 00:19:52.350 align:middle line:84% So this is one common way of how we actually 00:19:52.350 --> 00:19:54.210 align:middle line:90% like to look at these genes. 00:19:54.210 --> 00:19:56.460 align:middle line:84% It's called a heat map, or even a dendrogram, 00:19:56.460 --> 00:19:59.100 align:middle line:84% depending on the different programs that you use. 00:19:59.100 --> 00:20:01.530 align:middle line:84% But the green indicates-- so each of the rows 00:20:01.530 --> 00:20:02.610 align:middle line:90% indicates a gene. 00:20:02.610 --> 00:20:05.040 align:middle line:84% Each of the columns typically represents a sample. 00:20:05.040 --> 00:20:08.400 align:middle line:84% And the green indicates a gene that's higher than the mean, 00:20:08.400 --> 00:20:11.490 align:middle line:84% and the red indicates a gene that's lower than the mean. 00:20:11.490 --> 00:20:15.240 align:middle line:84% And what we're showing here are a subset of those 55 genes 00:20:15.240 --> 00:20:19.020 align:middle line:84% that actually have something to do with energy metabolism. 00:20:19.020 --> 00:20:21.300 align:middle line:84% And primarily most of the genes that 00:20:21.300 --> 00:20:23.460 align:middle line:84% have something to do with energy metabolism 00:20:23.460 --> 00:20:26.010 align:middle line:84% are higher in the family history negative, 00:20:26.010 --> 00:20:28.620 align:middle line:84% they're lower in the diabetics, and they're already 00:20:28.620 --> 00:20:30.390 align:middle line:84% low in the family history positive, 00:20:30.390 --> 00:20:32.475 align:middle line:84% so the ones that are at risk for getting diabetes. 00:20:32.475 --> 00:20:35.440 align:middle line:90% 00:20:35.440 --> 00:20:38.020 align:middle line:84% So now we've got a piece of this puzzle. 00:20:38.020 --> 00:20:39.960 align:middle line:84% We see that transcription of some 00:20:39.960 --> 00:20:42.690 align:middle line:84% of these mitochondrial genes, metabolic related genes, 00:20:42.690 --> 00:20:44.310 align:middle line:90% is down. 00:20:44.310 --> 00:20:46.810 align:middle line:84% That might be causing decreased oxidative phosphorylation. 00:20:46.810 --> 00:20:49.200 align:middle line:84% So what happens is, fat is building up 00:20:49.200 --> 00:20:51.600 align:middle line:90% in the muscle and the fat cells. 00:20:51.600 --> 00:20:53.400 align:middle line:84% And if you have less of these genes 00:20:53.400 --> 00:20:54.960 align:middle line:84% and presumably less of the proteins 00:20:54.960 --> 00:20:56.910 align:middle line:84% that we're using them as a proxy for, 00:20:56.910 --> 00:20:59.730 align:middle line:84% we're getting decreased lipid oxidation. 00:20:59.730 --> 00:21:01.800 align:middle line:84% As fat or lipid builds up in muscle, 00:21:01.800 --> 00:21:04.295 align:middle line:84% it actually can cause the insulin to not work as well. 00:21:04.295 --> 00:21:05.670 align:middle line:84% So maybe that's how we're getting 00:21:05.670 --> 00:21:07.800 align:middle line:84% to insulin resistance because of an increase 00:21:07.800 --> 00:21:09.510 align:middle line:90% in the lipid in the muscle. 00:21:09.510 --> 00:21:13.710 align:middle line:84% But there's a huge half of this that's still unknown today, 00:21:13.710 --> 00:21:17.640 align:middle line:84% and that's the part I want to focus on for a second. 00:21:17.640 --> 00:21:19.260 align:middle line:84% People show this slide all the time 00:21:19.260 --> 00:21:22.140 align:middle line:84% in all sorts of different contexts. 00:21:22.140 --> 00:21:25.050 align:middle line:84% Huge formula here, huge formula here, the miracle 00:21:25.050 --> 00:21:26.670 align:middle line:90% occurs in the middle. 00:21:26.670 --> 00:21:30.600 align:middle line:84% And still the biggest problem with this type of study 00:21:30.600 --> 00:21:34.080 align:middle line:84% is that this miracle still has to occur to figure out why 00:21:34.080 --> 00:21:36.640 align:middle line:90% do we have this list of genes. 00:21:36.640 --> 00:21:40.860 align:middle line:84% What is it about this list of genes that's special? 00:21:40.860 --> 00:21:43.930 align:middle line:84% How are we getting to this set of genes? 00:21:43.930 --> 00:21:45.940 align:middle line:90% What's the biology behind it? 00:21:45.940 --> 00:21:48.750 align:middle line:84% So although many of my colleagues will argue with me 00:21:48.750 --> 00:21:52.620 align:middle line:84% about this, you cannot ignore the biology here. 00:21:52.620 --> 00:21:54.430 align:middle line:84% It's getting harder and harder to just say, 00:21:54.430 --> 00:21:57.095 align:middle line:84% here's my list of genes, go run with this. 00:21:57.095 --> 00:21:59.220 align:middle line:84% You need to know what is it about this list, what's 00:21:59.220 --> 00:22:01.770 align:middle line:84% causing this list of genes to be so special. 00:22:01.770 --> 00:22:04.460 align:middle line:84% And unfortunately right now to informaticists 00:22:04.460 --> 00:22:06.960 align:middle line:84% it looks like a miracle occurs, that someone has the insight 00:22:06.960 --> 00:22:08.730 align:middle line:84% to look at one particular factor, 00:22:08.730 --> 00:22:11.010 align:middle line:84% and you actually get a causal chain of events, 00:22:11.010 --> 00:22:12.720 align:middle line:84% and then you can get your list of genes. 00:22:12.720 --> 00:22:15.750 align:middle line:84% Here what we realized is that most of these genes 00:22:15.750 --> 00:22:18.060 align:middle line:84% are downstream of a transcription factor called 00:22:18.060 --> 00:22:22.560 align:middle line:84% NRF-1 one and another one called PGC-1alpha. 00:22:22.560 --> 00:22:25.020 align:middle line:84% It'd be great if we had some database of these things. 00:22:25.020 --> 00:22:26.370 align:middle line:90% Of course, we don't. 00:22:26.370 --> 00:22:28.960 align:middle line:84% It'd be great if NRF-1 was even on the chip. 00:22:28.960 --> 00:22:30.090 align:middle line:90% It's not. 00:22:30.090 --> 00:22:32.370 align:middle line:84% It's on today's chip, but it wasn't on these chips 00:22:32.370 --> 00:22:34.290 align:middle line:90% from three generations ago. 00:22:34.290 --> 00:22:37.200 align:middle line:84% So no amount of informatics and network building 00:22:37.200 --> 00:22:38.640 align:middle line:84% and Bayes networks or any of that 00:22:38.640 --> 00:22:41.190 align:middle line:84% would have got us to this particular hypothesis, 00:22:41.190 --> 00:22:44.080 align:middle line:84% because the gene wasn't even being measured at the time. 00:22:44.080 --> 00:22:46.300 align:middle line:84% But all of these genes that I'm showing here-- again, 00:22:46.300 --> 00:22:48.000 align:middle line:84% a subset of the previous subset-- 00:22:48.000 --> 00:22:50.250 align:middle line:90% are all downstream of NRF-1. 00:22:50.250 --> 00:22:52.830 align:middle line:84% So that led us to actually think about the hypothesis, 00:22:52.830 --> 00:22:55.260 align:middle line:84% maybe NRF-1 is down, and that's why 00:22:55.260 --> 00:22:56.712 align:middle line:90% all these other genes are down. 00:22:56.712 --> 00:22:58.170 align:middle line:84% Because it's much easier to come up 00:22:58.170 --> 00:23:00.300 align:middle line:84% with a causal explanation requiring just one 00:23:00.300 --> 00:23:03.270 align:middle line:84% thing to be wrong than 50 things to be wrong. 00:23:03.270 --> 00:23:05.550 align:middle line:90% That's Occam's razor. 00:23:05.550 --> 00:23:07.210 align:middle line:90% Simpler explanations. 00:23:07.210 --> 00:23:09.090 align:middle line:90% So when you look at NRF levels-- 00:23:09.090 --> 00:23:12.060 align:middle line:84% again this measured a different population using 00:23:12.060 --> 00:23:14.738 align:middle line:90% a different technology, RT-PCR. 00:23:14.738 --> 00:23:16.530 align:middle line:84% You have the family history negatives here, 00:23:16.530 --> 00:23:18.000 align:middle line:90% positives here, diabetics again. 00:23:18.000 --> 00:23:21.990 align:middle line:84% And the gene actually has a statistically significant drop 00:23:21.990 --> 00:23:23.850 align:middle line:90% in the diabetics. 00:23:23.850 --> 00:23:25.350 align:middle line:84% But what's puzzling is that it's not 00:23:25.350 --> 00:23:26.400 align:middle line:90% dropping in the middle group. 00:23:26.400 --> 00:23:28.608 align:middle line:84% But we saw those genes were down in the middle group, 00:23:28.608 --> 00:23:32.070 align:middle line:84% so this isn't a perfect explanation at all. 00:23:32.070 --> 00:23:33.750 align:middle line:84% Another piece of biological knowledge 00:23:33.750 --> 00:23:37.200 align:middle line:84% led us to realize that NRF is actually itself is downstream, 00:23:37.200 --> 00:23:39.820 align:middle line:84% or actually it's co-activated by PGC-1alpha. 00:23:39.820 --> 00:23:41.340 align:middle line:84% So when you actually measure that, 00:23:41.340 --> 00:23:43.560 align:middle line:84% you also get the clue that PGC-1alpha 00:23:43.560 --> 00:23:45.750 align:middle line:84% is down in both of these two groups, 00:23:45.750 --> 00:23:47.730 align:middle line:84% as well as another form, PGC-1beta, which 00:23:47.730 --> 00:23:49.950 align:middle line:90% is down in these two groups. 00:23:49.950 --> 00:23:52.080 align:middle line:90% The biggest problem, though-- 00:23:52.080 --> 00:23:56.370 align:middle line:84% so now the hypothesis is that drops in PGC-1alpha and NRF 00:23:56.370 --> 00:23:59.532 align:middle line:84% are actually what's causing the difference between diabetics 00:23:59.532 --> 00:24:00.990 align:middle line:84% and non-diabetics, and it's already 00:24:00.990 --> 00:24:04.500 align:middle line:84% showing a difference in the pre-diabetics. 00:24:04.500 --> 00:24:07.260 align:middle line:84% The biggest problem, though, with this study 00:24:07.260 --> 00:24:11.315 align:middle line:84% is that what's the cause, and what's the effect here? 00:24:11.315 --> 00:24:14.310 align:middle line:84% I'm already showing you that the glucose levels and insulin 00:24:14.310 --> 00:24:16.620 align:middle line:84% levels are already high in that middle group. 00:24:16.620 --> 00:24:20.010 align:middle line:84% How do I know this isn't just an effect of having high sugar 00:24:20.010 --> 00:24:21.070 align:middle line:90% levels? 00:24:21.070 --> 00:24:23.880 align:middle line:84% How do I know this is causing the high sugar levels? 00:24:23.880 --> 00:24:27.090 align:middle line:84% And that's the problem with this type of approach. 00:24:27.090 --> 00:24:30.660 align:middle line:90% We can easily get samples now. 00:24:30.660 --> 00:24:33.010 align:middle line:90% Samples aren't the hard part. 00:24:33.010 --> 00:24:35.500 align:middle line:84% The arrays are not the hard part. 00:24:35.500 --> 00:24:37.355 align:middle line:84% The analysis is not the hard part. 00:24:37.355 --> 00:24:38.730 align:middle line:84% Informatics is not the hard part. 00:24:38.730 --> 00:24:41.220 align:middle line:84% There's plenty of people around now, plenty of software 00:24:41.220 --> 00:24:42.900 align:middle line:90% you can download. 00:24:42.900 --> 00:24:47.310 align:middle line:84% The hard part's the interpretation in the end. 00:24:47.310 --> 00:24:50.310 align:middle line:84% Even though we have all of these genome scale tools, 00:24:50.310 --> 00:24:53.490 align:middle line:84% we still have nothing that makes the interpretation that much 00:24:53.490 --> 00:24:55.900 align:middle line:90% faster or that much easier. 00:24:55.900 --> 00:24:57.060 align:middle line:90% So now it's still routine. 00:24:57.060 --> 00:24:58.643 align:middle line:84% If I go to the Joslin Diabetes Center, 00:24:58.643 --> 00:25:00.210 align:middle line:90% I look at their core facility. 00:25:00.210 --> 00:25:01.800 align:middle line:84% I was just there a couple of days ago. 00:25:01.800 --> 00:25:03.960 align:middle line:84% They've literally built a pyramid 00:25:03.960 --> 00:25:07.800 align:middle line:84% of all the arrays they've done so far, on the order of about 00:25:07.800 --> 00:25:10.800 align:middle line:84% 1,500 almost 2,000 microarrays already done 00:25:10.800 --> 00:25:13.470 align:middle line:90% at the Joslin Diabetes Center. 00:25:13.470 --> 00:25:17.310 align:middle line:84% There's maybe been about two papers published so far 00:25:17.310 --> 00:25:20.280 align:middle line:84% with all of this array data, this one which 00:25:20.280 --> 00:25:22.170 align:middle line:84% had about 30 samples, and another one 00:25:22.170 --> 00:25:23.730 align:middle line:90% that had about 10 samples. 00:25:23.730 --> 00:25:26.190 align:middle line:84% And still there are thousands of arrays that 00:25:26.190 --> 00:25:27.480 align:middle line:90% have already been collected. 00:25:27.480 --> 00:25:29.910 align:middle line:84% The data is already sitting there. 00:25:29.910 --> 00:25:31.300 align:middle line:90% Analysis has already been done. 00:25:31.300 --> 00:25:33.900 align:middle line:84% But people cannot make the leap to the next step. 00:25:33.900 --> 00:25:36.000 align:middle line:90% Why is it this list of genes? 00:25:36.000 --> 00:25:38.520 align:middle line:84% What's so interesting about that list of genes? 00:25:38.520 --> 00:25:41.100 align:middle line:84% I think that's going to be the hard part now 00:25:41.100 --> 00:25:42.550 align:middle line:90% in the next few years. 00:25:42.550 --> 00:25:47.100 align:middle line:84% This is not that novel to take samples and put them on arrays. 00:25:47.100 --> 00:25:48.960 align:middle line:84% That's not what the medicine part of this 00:25:48.960 --> 00:25:50.080 align:middle line:90% is going to be about. 00:25:50.080 --> 00:25:51.538 align:middle line:84% It's going to be trying to get down 00:25:51.538 --> 00:25:53.760 align:middle line:84% to simpler explanations for what we're seeing here. 00:25:53.760 --> 00:25:57.590 align:middle line:90% 00:25:57.590 --> 00:25:59.410 align:middle line:84% So this is a final hypothesis here, 00:25:59.410 --> 00:26:03.130 align:middle line:84% perhaps that PGC-1 is down, and that, along with NRF, 00:26:03.130 --> 00:26:05.560 align:middle line:84% is actually causing this list of genes to be different. 00:26:05.560 --> 00:26:08.677 align:middle line:90% 00:26:08.677 --> 00:26:10.010 align:middle line:90% Any questions about that so far? 00:26:10.010 --> 00:26:12.930 align:middle line:90% Any thoughts? 00:26:12.930 --> 00:26:15.840 align:middle line:90% AUDIENCE: Do you get any-- 00:26:15.840 --> 00:26:20.140 align:middle line:90% 00:26:20.140 --> 00:26:22.120 align:middle line:84% we look at that AMP kinase pathway 00:26:22.120 --> 00:26:24.702 align:middle line:90% activation in terms of toxins? 00:26:24.702 --> 00:26:25.910 align:middle line:90% ATUL BUTTE: A great question. 00:26:25.910 --> 00:26:28.540 align:middle line:84% So the question is, have we looked at the AMP kinase 00:26:28.540 --> 00:26:29.390 align:middle line:90% pathway? 00:26:29.390 --> 00:26:32.710 align:middle line:84% So we know, for example, when people exercise, 00:26:32.710 --> 00:26:35.920 align:middle line:84% their sensitivity to insulin improves, 00:26:35.920 --> 00:26:37.660 align:middle line:90% even if they have diabetes. 00:26:37.660 --> 00:26:40.390 align:middle line:84% People who exercise, the same molecule of insulin 00:26:40.390 --> 00:26:42.200 align:middle line:90% has more of an effect. 00:26:42.200 --> 00:26:44.470 align:middle line:84% And that's mediated through a different pathway, 00:26:44.470 --> 00:26:46.900 align:middle line:90% through the AMP kinase pathway. 00:26:46.900 --> 00:26:49.060 align:middle line:84% Did we look at AMP kinase levels here? 00:26:49.060 --> 00:26:49.600 align:middle line:90% No. 00:26:49.600 --> 00:26:51.190 align:middle line:84% Did we look at the expression levels 00:26:51.190 --> 00:26:53.230 align:middle line:84% that might have been on the chip but they weren't different? 00:26:53.230 --> 00:26:55.272 align:middle line:84% Did we look at activity levels or protein levels? 00:26:55.272 --> 00:26:57.010 align:middle line:90% No. 00:26:57.010 --> 00:27:00.250 align:middle line:84% But you can bet in that towering pile of 1,000 arrays-- 00:27:00.250 --> 00:27:03.670 align:middle line:84% Lori Goodyear and others have done these arrays already-- 00:27:03.670 --> 00:27:06.010 align:middle line:84% but they're still stuck the same way. 00:27:06.010 --> 00:27:10.150 align:middle line:84% So what is it about these lists of genes that's going 00:27:10.150 --> 00:27:13.720 align:middle line:90% to get us to a hypothesis here? 00:27:13.720 --> 00:27:15.340 align:middle line:90% It's very unwieldy. 00:27:15.340 --> 00:27:17.620 align:middle line:84% It's very hard to deal with these lists of hundreds 00:27:17.620 --> 00:27:19.660 align:middle line:90% of genes. 00:27:19.660 --> 00:27:21.490 align:middle line:84% It's not very validating anymore. 00:27:21.490 --> 00:27:23.770 align:middle line:84% It's not very fun to work with these genes. 00:27:23.770 --> 00:27:26.680 align:middle line:84% The hard part is to get down to the causal mechanism 00:27:26.680 --> 00:27:29.350 align:middle line:84% for those genes, and that's where we have the least 00:27:29.350 --> 00:27:32.420 align:middle line:84% information right now, like what's upstream of what, 00:27:32.420 --> 00:27:34.190 align:middle line:90% for example. 00:27:34.190 --> 00:27:36.730 align:middle line:84% AUDIENCE: Can you go back to the chicken and egg 00:27:36.730 --> 00:27:42.640 align:middle line:84% conundrum [INAUDIBLE] I was just wondering 00:27:42.640 --> 00:27:46.870 align:middle line:84% how do you sustain a direct decrease in ox pause, 00:27:46.870 --> 00:27:47.990 align:middle line:90% lipid oxidation. 00:27:47.990 --> 00:27:55.770 align:middle line:84% So the idea is that you're not using your glucose efficiently. 00:27:55.770 --> 00:27:57.960 align:middle line:90% when you blood sugar is high. 00:27:57.960 --> 00:28:00.464 align:middle line:90% ATUL BUTTE: Exactly. 00:28:00.464 --> 00:28:05.340 align:middle line:84% AUDIENCE: You're not metabolizing [INAUDIBLE] 00:28:05.340 --> 00:28:06.267 align:middle line:90% 00:28:06.267 --> 00:28:08.850 align:middle line:84% ATUL BUTTE: Primarily you're not generating ATP as efficiently 00:28:08.850 --> 00:28:12.120 align:middle line:90% as you could in a muscle. 00:28:12.120 --> 00:28:17.850 align:middle line:84% AUDIENCE: Yeah, well if you're generating ATP inefficiently, 00:28:17.850 --> 00:28:20.580 align:middle line:84% and that's the equivalent of having off proxy coupling 00:28:20.580 --> 00:28:21.720 align:middle line:90% or something. 00:28:21.720 --> 00:28:22.980 align:middle line:90% [INTERPOSING VOICES] 00:28:22.980 --> 00:28:23.940 align:middle line:90% ATUL BUTTE: Exactly. 00:28:23.940 --> 00:28:26.790 align:middle line:84% AUDIENCE: This is you're actually not generating the ATP 00:28:26.790 --> 00:28:27.450 align:middle line:90% and instead-- 00:28:27.450 --> 00:28:28.350 align:middle line:90% ATUL BUTTE: Yeah, the pipe is-- 00:28:28.350 --> 00:28:28.770 align:middle line:90% AUDIENCE: --substrate. 00:28:28.770 --> 00:28:29.603 align:middle line:90% ATUL BUTTE: Exactly. 00:28:29.603 --> 00:28:31.290 align:middle line:84% The pipe is not being filled, exactly. 00:28:31.290 --> 00:28:33.840 align:middle line:84% The pipe is not being filled as efficiently as it can be. 00:28:33.840 --> 00:28:34.532 align:middle line:90% AUDIENCE: Right. 00:28:34.532 --> 00:28:35.490 align:middle line:90% ATUL BUTTE: Absolutely. 00:28:35.490 --> 00:28:37.115 align:middle line:84% AUDIENCE: If we use that and we go back 00:28:37.115 --> 00:28:39.900 align:middle line:84% maybe to sort of backwards reasoning, 00:28:39.900 --> 00:28:44.700 align:middle line:84% wouldn't that basically suggest that the initial perturbation 00:28:44.700 --> 00:28:48.326 align:middle line:90% is in NRF and CTAR? 00:28:48.326 --> 00:28:50.795 align:middle line:90% 00:28:50.795 --> 00:28:52.170 align:middle line:84% ATUL BUTTE: So there's other ways 00:28:52.170 --> 00:28:54.090 align:middle line:84% we can-- there's another piece of information 00:28:54.090 --> 00:28:56.280 align:middle line:84% that we can bring to try to figure out 00:28:56.280 --> 00:28:57.780 align:middle line:90% the ordering of them. 00:28:57.780 --> 00:29:00.420 align:middle line:84% So there are sequenced polymorphisms 00:29:00.420 --> 00:29:03.540 align:middle line:84% that are known in these genes, specifically 00:29:03.540 --> 00:29:07.530 align:middle line:84% in PGC-1, that go along to having diabetes. 00:29:07.530 --> 00:29:09.640 align:middle line:84% So we think, well, if there's people 00:29:09.640 --> 00:29:12.510 align:middle line:84% who have an abnormal sequence and they get diabetes, 00:29:12.510 --> 00:29:14.702 align:middle line:90% maybe that's what's causing it. 00:29:14.702 --> 00:29:15.660 align:middle line:90% AUDIENCE: Right, right. 00:29:15.660 --> 00:29:17.910 align:middle line:84% ATUL BUTTE: That's what puts this upstream of actually 00:29:17.910 --> 00:29:19.440 align:middle line:90% having the high sugar levels. 00:29:19.440 --> 00:29:22.560 align:middle line:84% But in this sample, without that a priori knowledge, 00:29:22.560 --> 00:29:23.730 align:middle line:90% you can't tell that. 00:29:23.730 --> 00:29:24.838 align:middle line:90% That's the issue. 00:29:24.838 --> 00:29:26.880 align:middle line:84% AUDIENCE: Doesn't it suggest that, though, or was 00:29:26.880 --> 00:29:28.230 align:middle line:90% that a giant leap? 00:29:28.230 --> 00:29:31.740 align:middle line:84% In other words, if the glucose level was high, 00:29:31.740 --> 00:29:36.450 align:middle line:84% if the initial perturbation was the elevated blood glucose-- 00:29:36.450 --> 00:29:39.420 align:middle line:84% ATUL BUTTE: OK, yep, if it were that. 00:29:39.420 --> 00:29:47.000 align:middle line:84% AUDIENCE: --that doesn't explain why you have upregulation of-- 00:29:47.000 --> 00:29:51.200 align:middle line:84% or is it downregulation of NRF and PGC, whatever it is. 00:29:51.200 --> 00:29:52.340 align:middle line:90% It up or down? 00:29:52.340 --> 00:29:53.570 align:middle line:90% ATUL BUTTE: The NRF is down. 00:29:53.570 --> 00:29:56.090 align:middle line:84% All I'm saying is with just this experiment, 00:29:56.090 --> 00:30:00.320 align:middle line:84% I could put this square up here and say 00:30:00.320 --> 00:30:02.810 align:middle line:84% that is leading to this, leading to this, leading to this, 00:30:02.810 --> 00:30:04.610 align:middle line:90% leading to this. 00:30:04.610 --> 00:30:08.097 align:middle line:84% I can't make that leap that this is causal with this. 00:30:08.097 --> 00:30:08.930 align:middle line:90% [INTERPOSING VOICES] 00:30:08.930 --> 00:30:11.150 align:middle line:84% AUDIENCE: More suggested that the square's down there, 00:30:11.150 --> 00:30:12.380 align:middle line:84% or am I not thinking about this correctly? 00:30:12.380 --> 00:30:14.120 align:middle line:84% ATUL BUTTE: With the sequenced polymorphisms and things, 00:30:14.120 --> 00:30:15.980 align:middle line:84% it makes me think that it's down here. 00:30:15.980 --> 00:30:17.990 align:middle line:84% But just the experiment by itself 00:30:17.990 --> 00:30:20.420 align:middle line:84% without the a priori knowledge, without knowing 00:30:20.420 --> 00:30:22.955 align:middle line:84% what the literature tells you, how would 00:30:22.955 --> 00:30:24.830 align:middle line:84% you know that, because these patients already 00:30:24.830 --> 00:30:27.290 align:middle line:90% have some degree of resistance. 00:30:27.290 --> 00:30:28.730 align:middle line:90% That's the issue here. 00:30:28.730 --> 00:30:30.290 align:middle line:84% You need to know what-- the biology. 00:30:30.290 --> 00:30:32.180 align:middle line:90% You cannot ignore the biology. 00:30:32.180 --> 00:30:35.090 align:middle line:84% AUDIENCE: Right, but the question is just the-- 00:30:35.090 --> 00:30:39.080 align:middle line:84% [INAUDIBLE] Because if you look at the control group, 00:30:39.080 --> 00:30:40.987 align:middle line:84% the body mass index seem to be much lower. 00:30:40.987 --> 00:30:41.945 align:middle line:90% ATUL BUTTE: Absolutely. 00:30:41.945 --> 00:30:43.010 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:30:43.010 --> 00:30:44.360 align:middle line:90% ATUL BUTTE: Exactly right. 00:30:44.360 --> 00:30:47.480 align:middle line:84% So you picked up on that as well? 00:30:47.480 --> 00:30:49.160 align:middle line:84% The body mass index is different, too, 00:30:49.160 --> 00:30:50.990 align:middle line:90% because it's hard to find. 00:30:50.990 --> 00:30:52.880 align:middle line:84% I mean, that's life in America now. 00:30:52.880 --> 00:30:54.680 align:middle line:90% Everyone is obese. 00:30:54.680 --> 00:30:58.100 align:middle line:84% And so even teenagers that are obese 00:30:58.100 --> 00:30:59.922 align:middle line:84% might be in this middle group now. 00:30:59.922 --> 00:31:01.880 align:middle line:84% They will have some impaired glucose tolerance. 00:31:01.880 --> 00:31:03.860 align:middle line:84% AUDIENCE: Of this case in the control group, 00:31:03.860 --> 00:31:06.710 align:middle line:90% control group was normal. 00:31:06.710 --> 00:31:08.420 align:middle line:84% ATUL BUTTE: The control group was-- 00:31:08.420 --> 00:31:11.690 align:middle line:84% I mean, their body mass index is still 30. 00:31:11.690 --> 00:31:14.030 align:middle line:90% That qualifies as obese today. 00:31:14.030 --> 00:31:16.520 align:middle line:84% And the family history group is actually slightly less. 00:31:16.520 --> 00:31:18.925 align:middle line:84% But they're all obese, basically, 00:31:18.925 --> 00:31:19.925 align:middle line:90% especially the controls. 00:31:19.925 --> 00:31:21.740 align:middle line:84% AUDIENCE: But you can see the distribution 00:31:21.740 --> 00:31:23.032 align:middle line:90% on the graphs on the next page. 00:31:23.032 --> 00:31:26.090 align:middle line:84% ATUL BUTTE: Absolutely, absolutely. 00:31:26.090 --> 00:31:26.767 align:middle line:90% This one here. 00:31:26.767 --> 00:31:27.350 align:middle line:90% AUDIENCE: Yep. 00:31:27.350 --> 00:31:29.210 align:middle line:90% ATUL BUTTE: Yep, exactly. 00:31:29.210 --> 00:31:31.050 align:middle line:84% This is one gene versus the other, 00:31:31.050 --> 00:31:33.260 align:middle line:84% and you can see the body mass index here, exactly. 00:31:33.260 --> 00:31:35.690 align:middle line:84% And you got one all the way over here, exactly. 00:31:35.690 --> 00:31:37.380 align:middle line:90% That's destroying the mean. 00:31:37.380 --> 00:31:39.320 align:middle line:84% So these are not otherwise equal groups. 00:31:39.320 --> 00:31:41.030 align:middle line:90% That's the problem. 00:31:41.030 --> 00:31:43.760 align:middle line:84% Now you have to-- so why am I even showing all this, 00:31:43.760 --> 00:31:46.730 align:middle line:84% if I have all these problems with this analysis 00:31:46.730 --> 00:31:47.737 align:middle line:90% and this data? 00:31:47.737 --> 00:31:49.320 align:middle line:84% The whole point of this kind of course 00:31:49.320 --> 00:31:53.972 align:middle line:84% is not just to teach you how to do these types of experiments. 00:31:53.972 --> 00:31:55.430 align:middle line:84% To me it would be great if you knew 00:31:55.430 --> 00:31:57.680 align:middle line:84% how to interpret these things that come out in the New 00:31:57.680 --> 00:31:59.900 align:middle line:90% England Journal every day. 00:31:59.900 --> 00:32:02.330 align:middle line:84% Because every month now we see a microarray paper, say, 00:32:02.330 --> 00:32:04.850 align:middle line:84% in the New England Journal, or JAMA, or other equivalent 00:32:04.850 --> 00:32:06.140 align:middle line:90% publications. 00:32:06.140 --> 00:32:07.700 align:middle line:84% And I think we have to teach people 00:32:07.700 --> 00:32:10.700 align:middle line:84% how to read these things with a critical eye. 00:32:10.700 --> 00:32:12.560 align:middle line:84% Because it looks so fancy to come out 00:32:12.560 --> 00:32:14.390 align:middle line:84% with 44,000 genes, and these are the ones 00:32:14.390 --> 00:32:15.920 align:middle line:90% that are actually diagnostic. 00:32:15.920 --> 00:32:18.560 align:middle line:84% But there's actually problems with the experimental design. 00:32:18.560 --> 00:32:21.260 align:middle line:84% And if the experiment was not designed in exactly the way 00:32:21.260 --> 00:32:23.330 align:middle line:84% to answer the question that was being asked, 00:32:23.330 --> 00:32:26.090 align:middle line:84% no amount of fancy arrays is going to be able to help you. 00:32:26.090 --> 00:32:28.418 align:middle line:90% You can't salvage that. 00:32:28.418 --> 00:32:29.960 align:middle line:84% That's what I'm trying to get across, 00:32:29.960 --> 00:32:33.410 align:middle line:90% is how to be critical here. 00:32:33.410 --> 00:32:34.513 align:middle line:90% I think the same thing-- 00:32:34.513 --> 00:32:35.930 align:middle line:84% when you see a bunch of genes that 00:32:35.930 --> 00:32:39.740 align:middle line:90% distinguishes ALL from AML. 00:32:39.740 --> 00:32:42.110 align:middle line:84% Let's think about all the other things in their context 00:32:42.110 --> 00:32:44.990 align:middle line:84% that could be making a difference here. 00:32:44.990 --> 00:32:46.490 align:middle line:90% When were the samples acquired? 00:32:46.490 --> 00:32:47.810 align:middle line:90% Where were they acquired? 00:32:47.810 --> 00:32:50.450 align:middle line:84% Are patients with AML preferentially 00:32:50.450 --> 00:32:52.760 align:middle line:84% sent to one center, and the ones with ALL 00:32:52.760 --> 00:32:55.490 align:middle line:90% sent to another center? 00:32:55.490 --> 00:32:57.890 align:middle line:84% My favorite example I love to give is this. 00:32:57.890 --> 00:32:59.640 align:middle line:84% Let's say you're looking at a solid tumor, 00:32:59.640 --> 00:33:01.598 align:middle line:84% and you're trying to distinguish the genes that 00:33:01.598 --> 00:33:04.070 align:middle line:84% are different between the metastatic form of this cancer 00:33:04.070 --> 00:33:06.560 align:middle line:90% and the non-metastatic form. 00:33:06.560 --> 00:33:08.240 align:middle line:84% If you've spent any time in the wards, 00:33:08.240 --> 00:33:10.550 align:middle line:90% you know how surgeons work. 00:33:10.550 --> 00:33:12.680 align:middle line:84% If you have a patient, and if a surgeon is looking 00:33:12.680 --> 00:33:14.275 align:middle line:84% at their caseload for the next day, 00:33:14.275 --> 00:33:16.400 align:middle line:84% and they have a patient with metastatic and patient 00:33:16.400 --> 00:33:19.040 align:middle line:84% with non-metastatic, the patient metastatic, 00:33:19.040 --> 00:33:20.160 align:middle line:90% it's already metastasized. 00:33:20.160 --> 00:33:22.590 align:middle line:84% The surgery is going to be a hell of a lot longer, 00:33:22.590 --> 00:33:24.470 align:middle line:84% let's say, because they have to go, 00:33:24.470 --> 00:33:27.583 align:middle line:84% check this, check that, clean this out, clean that out. 00:33:27.583 --> 00:33:29.750 align:middle line:84% The surgeon might say, I'm going to put that patient 00:33:29.750 --> 00:33:32.477 align:middle line:90% as the first case of the day. 00:33:32.477 --> 00:33:34.310 align:middle line:84% And the simpler case, I'm going to put on as 00:33:34.310 --> 00:33:36.338 align:middle line:90% a follow-on or an add-on case. 00:33:36.338 --> 00:33:37.880 align:middle line:84% Keep them waiting in the waiting room 00:33:37.880 --> 00:33:39.570 align:middle line:84% or in the recovery room ready to go, 00:33:39.570 --> 00:33:42.440 align:middle line:84% and then we're going to do them after this first case is done. 00:33:42.440 --> 00:33:44.318 align:middle line:84% So now imagine a study where you look 00:33:44.318 --> 00:33:46.610 align:middle line:84% at the difference between metastatic and non-metastatic 00:33:46.610 --> 00:33:48.950 align:middle line:84% forms of this particular cancer, and you've 00:33:48.950 --> 00:33:52.370 align:middle line:84% got these samples from these surgeons who preferentially put 00:33:52.370 --> 00:33:55.370 align:middle line:84% the hard cases in the morning and the easy cases 00:33:55.370 --> 00:33:57.500 align:middle line:90% in the afternoon. 00:33:57.500 --> 00:33:58.610 align:middle line:90% I'm an endocrinologist. 00:33:58.610 --> 00:34:01.250 align:middle line:84% I know that are hormones that are different between morning 00:34:01.250 --> 00:34:05.060 align:middle line:84% and evening, like insulin, growth hormone, cortisol. 00:34:05.060 --> 00:34:09.409 align:middle line:84% Who knows how many genes are downstream of those hormones? 00:34:09.409 --> 00:34:11.750 align:middle line:84% How do you know that the signal that you're seeing 00:34:11.750 --> 00:34:15.810 align:middle line:84% isn't because of that bias or that confounding there? 00:34:15.810 --> 00:34:18.860 align:middle line:84% Now this is not revolutionary stuff. 00:34:18.860 --> 00:34:21.080 align:middle line:84% If we looked at clinical tests, the clinical trials 00:34:21.080 --> 00:34:23.610 align:middle line:84% in the past, we knew to look for these things. 00:34:23.610 --> 00:34:25.610 align:middle line:84% But all of a sudden now that we have these fancy 00:34:25.610 --> 00:34:27.920 align:middle line:84% red/green-ograms, and we have these microarrays, 00:34:27.920 --> 00:34:30.650 align:middle line:84% people have forgotten about all of this traditional-- 00:34:30.650 --> 00:34:33.350 align:middle line:90% these traditional confounders. 00:34:33.350 --> 00:34:36.050 align:middle line:84% We're measuring across the entire measurable genome. 00:34:36.050 --> 00:34:38.600 align:middle line:84% The signal that you get just from endocrinology, 00:34:38.600 --> 00:34:41.719 align:middle line:84% from circadian cycles, might be stronger than the signal 00:34:41.719 --> 00:34:44.270 align:middle line:84% that you actually think you're measuring. 00:34:44.270 --> 00:34:47.062 align:middle line:84% Now you go back and look at the last three microarray papers 00:34:47.062 --> 00:34:49.020 align:middle line:84% that are in the New England Journal of Medicine 00:34:49.020 --> 00:34:51.780 align:middle line:84% and just look to see if you can figure out what time of day 00:34:51.780 --> 00:34:53.310 align:middle line:90% these samples were acquired. 00:34:53.310 --> 00:34:55.380 align:middle line:90% No way. 00:34:55.380 --> 00:34:58.260 align:middle line:84% AUDIENCE: How sensitive is something like that? 00:34:58.260 --> 00:34:59.370 align:middle line:90% Well, two-part question. 00:34:59.370 --> 00:35:03.570 align:middle line:84% One, is it a matter of is there an 8:00 AM, an 8:30, 00:35:03.570 --> 00:35:04.710 align:middle line:90% and that sort of thing. 00:35:04.710 --> 00:35:07.620 align:middle line:90% And is that even known? 00:35:07.620 --> 00:35:08.370 align:middle line:90% ATUL BUTTE: Right. 00:35:08.370 --> 00:35:10.920 align:middle line:84% So it might not be known, but then 00:35:10.920 --> 00:35:15.870 align:middle line:84% what I would say is we should randomize across that variable. 00:35:15.870 --> 00:35:18.840 align:middle line:84% You could turn the lights on there. 00:35:18.840 --> 00:35:21.450 align:middle line:84% We should randomize across that variable. 00:35:21.450 --> 00:35:23.940 align:middle line:84% So for example, someone should be looking at the clock 00:35:23.940 --> 00:35:25.890 align:middle line:84% and say, well, these are all acquired-- 00:35:25.890 --> 00:35:27.360 align:middle line:84% there's some that are here, some that are here, 00:35:27.360 --> 00:35:29.860 align:middle line:84% some that are here, and just convince yourself that it's not 00:35:29.860 --> 00:35:31.740 align:middle line:90% just a gross bias. 00:35:31.740 --> 00:35:35.160 align:middle line:84% I'm just making a fake story here that's believable. 00:35:35.160 --> 00:35:37.500 align:middle line:84% AUDIENCE: Couldn't you just do normal controls 00:35:37.500 --> 00:35:40.032 align:middle line:84% and do a raise on them in the morning and the afternoon? 00:35:40.032 --> 00:35:40.740 align:middle line:90% ATUL BUTTE: Sure. 00:35:40.740 --> 00:35:44.190 align:middle line:84% You can try to compensate for some other way. 00:35:44.190 --> 00:35:46.440 align:middle line:84% But it's going to be relative to their hormone levels, 00:35:46.440 --> 00:35:47.850 align:middle line:90% too, probably. 00:35:47.850 --> 00:35:50.580 align:middle line:84% AUDIENCE: I guess the example that you gave and just 00:35:50.580 --> 00:35:56.400 align:middle line:84% going out there, but I guess it's 00:35:56.400 --> 00:35:58.110 align:middle line:84% different than in a laboratory setting, 00:35:58.110 --> 00:36:02.840 align:middle line:84% when [INTERPOSING VOICES] cells and eventually they die. 00:36:02.840 --> 00:36:04.930 align:middle line:90% ATUL BUTTE: Exactly. 00:36:04.930 --> 00:36:05.430 align:middle line:90% Exactly. 00:36:05.430 --> 00:36:08.340 align:middle line:84% We've taken this technology that we've used in micro-- 00:36:08.340 --> 00:36:11.530 align:middle line:84% used in cells, in cell culture from cell lines 00:36:11.530 --> 00:36:17.040 align:middle line:84% and rapidly moved it to patients that exist in a context. 00:36:17.040 --> 00:36:20.557 align:middle line:84% Even the middle ground of just using them in lab mice, 00:36:20.557 --> 00:36:22.890 align:middle line:84% look at the same mice, same conditions, same everything. 00:36:22.890 --> 00:36:25.230 align:middle line:84% There's genes that are different. 00:36:25.230 --> 00:36:28.050 align:middle line:84% One of the mice was an alpha male in that little cage, 00:36:28.050 --> 00:36:28.710 align:middle line:90% and one wasn't. 00:36:28.710 --> 00:36:30.420 align:middle line:84% There's differences between things 00:36:30.420 --> 00:36:32.070 align:middle line:84% that we think are otherwise the same. 00:36:32.070 --> 00:36:33.612 align:middle line:84% But now we figure it to this extreme, 00:36:33.612 --> 00:36:35.737 align:middle line:84% because there's a course, you're in a course called 00:36:35.737 --> 00:36:38.100 align:middle line:84% Genomic Medicine, and that's what we're doing today. 00:36:38.100 --> 00:36:39.960 align:middle line:84% These are the problems that we have 00:36:39.960 --> 00:36:42.630 align:middle line:90% with that type of approach. 00:36:42.630 --> 00:36:43.673 align:middle line:90% I kid you not. 00:36:43.673 --> 00:36:45.840 align:middle line:84% Go back to the last few microarray papers in the New 00:36:45.840 --> 00:36:48.090 align:middle line:84% England Journal, you will not see what time of day, 00:36:48.090 --> 00:36:50.310 align:middle line:84% even if you go to the supplemental material. 00:36:50.310 --> 00:36:52.170 align:middle line:84% Nobody might not-- I think people don't even 00:36:52.170 --> 00:36:53.760 align:middle line:90% write that stuff down. 00:36:53.760 --> 00:36:54.900 align:middle line:90% And it's a problem. 00:36:54.900 --> 00:36:57.760 align:middle line:84% I've only seen one paper so far that's addressed this issue. 00:36:57.760 --> 00:36:59.385 align:middle line:84% AUDIENCE: So what would your suggestion 00:36:59.385 --> 00:37:01.230 align:middle line:90% be, I mean, obviously to us? 00:37:01.230 --> 00:37:02.610 align:middle line:90% ATUL BUTTE: Control for that. 00:37:02.610 --> 00:37:05.110 align:middle line:84% AUDIENCE: It's to control for it, but in the interim, 00:37:05.110 --> 00:37:07.620 align:middle line:84% while there's labs that maybe really 00:37:07.620 --> 00:37:08.970 align:middle line:90% aren't thinking about this. 00:37:08.970 --> 00:37:09.637 align:middle line:90% ATUL BUTTE: Yep. 00:37:09.637 --> 00:37:11.137 align:middle line:84% You can do what [INTERPOSING VOICES] 00:37:11.137 --> 00:37:11.820 align:middle line:90% Sunil is saying. 00:37:11.820 --> 00:37:13.195 align:middle line:84% AUDIENCE: Stop what you're doing. 00:37:13.195 --> 00:37:15.180 align:middle line:84% ATUL BUTTE: No, but you can try to model that. 00:37:15.180 --> 00:37:17.640 align:middle line:84% So take some other tissue that's related 00:37:17.640 --> 00:37:20.708 align:middle line:84% and see just across a sampling of time points 00:37:20.708 --> 00:37:23.250 align:middle line:84% what genes are different, and maybe you can then subtract out 00:37:23.250 --> 00:37:24.060 align:middle line:90% that effect. 00:37:24.060 --> 00:37:25.590 align:middle line:84% But more importantly what I would 00:37:25.590 --> 00:37:28.080 align:middle line:84% say is, if you're going to make a list of genes 00:37:28.080 --> 00:37:30.893 align:middle line:84% that you think are diagnostic, we're going to have to-- 00:37:30.893 --> 00:37:32.310 align:middle line:84% as physicians, we're going to have 00:37:32.310 --> 00:37:35.280 align:middle line:84% to look at that with a very skeptical eye. 00:37:35.280 --> 00:37:38.190 align:middle line:84% Because you cannot control for everything here. 00:37:38.190 --> 00:37:41.160 align:middle line:84% But we're measuring across so many genes 00:37:41.160 --> 00:37:45.900 align:middle line:84% and have so few samples, that no matter what, 00:37:45.900 --> 00:37:47.790 align:middle line:84% we're going to be overfitting the data here. 00:37:47.790 --> 00:37:50.410 align:middle line:90% That's what I'm worried about. 00:37:50.410 --> 00:37:53.100 align:middle line:84% So we'll be overfitting what we have. 00:37:53.100 --> 00:37:54.930 align:middle line:84% No amount of control samples is going 00:37:54.930 --> 00:37:57.030 align:middle line:84% to help us fix that problem, I think. 00:37:57.030 --> 00:38:00.110 align:middle line:90% 00:38:00.110 --> 00:38:01.250 align:middle line:90% How are we doing for time? 00:38:01.250 --> 00:38:01.750 align:middle line:90% OK. 00:38:01.750 --> 00:38:05.990 align:middle line:90% 00:38:05.990 --> 00:38:11.330 align:middle line:84% Let's talk-- let's go a little bit more into the diabetes. 00:38:11.330 --> 00:38:14.480 align:middle line:84% And so what I was planning to do is talk about one more diabetes 00:38:14.480 --> 00:38:16.670 align:middle line:90% project that we've been doing. 00:38:16.670 --> 00:38:19.370 align:middle line:84% Then I'm going to talk about some web-based resources 00:38:19.370 --> 00:38:22.610 align:middle line:84% for doing some interesting hypothesis generation, 00:38:22.610 --> 00:38:25.290 align:middle line:84% and we'll just end it at that point. 00:38:25.290 --> 00:38:28.280 align:middle line:84% So all of you are-- so not all of you 00:38:28.280 --> 00:38:30.950 align:middle line:84% are actually are physicians or medical students, 00:38:30.950 --> 00:38:33.630 align:middle line:84% but nonetheless I'll present this case of a patient 00:38:33.630 --> 00:38:35.210 align:middle line:90% that I saw. 00:38:35.210 --> 00:38:37.670 align:middle line:84% This was a female presented with Acanthosis nigricans, 00:38:37.670 --> 00:38:39.320 align:middle line:90% random glucose of 162. 00:38:39.320 --> 00:38:42.050 align:middle line:84% So Acanthosis nigricans, do you know what that is? 00:38:42.050 --> 00:38:43.490 align:middle line:84% There's a dark patch that you can 00:38:43.490 --> 00:38:45.890 align:middle line:84% get in the back of your neck, and your elbows, 00:38:45.890 --> 00:38:47.210 align:middle line:90% and your underarms. 00:38:47.210 --> 00:38:51.170 align:middle line:84% And what that means is that your insulin is too high. 00:38:51.170 --> 00:38:54.650 align:middle line:84% Because insulin is telling the melanocytes 00:38:54.650 --> 00:38:57.790 align:middle line:84% through some unknown mechanism, it's telling the melanocytes 00:38:57.790 --> 00:38:59.570 align:middle line:90% to become darker. 00:38:59.570 --> 00:39:01.843 align:middle line:84% So you just look at people on the street 00:39:01.843 --> 00:39:04.010 align:middle line:84% now, and look to see if they have a dark patch here. 00:39:04.010 --> 00:39:04.790 align:middle line:90% It's not dirt. 00:39:04.790 --> 00:39:06.027 align:middle line:90% You can't scrub it off. 00:39:06.027 --> 00:39:07.610 align:middle line:84% That's a sign of Acanthosis nigricans, 00:39:07.610 --> 00:39:09.790 align:middle line:84% so someone's having a problem with insulin. 00:39:09.790 --> 00:39:12.050 align:middle line:90% It's too high. 00:39:12.050 --> 00:39:15.320 align:middle line:84% Did more studies, the LDL is sky high. 00:39:15.320 --> 00:39:17.330 align:middle line:84% LDL is high, total cholesterol sky high, 00:39:17.330 --> 00:39:20.640 align:middle line:84% triglycerides are very high, HDL is low. 00:39:20.640 --> 00:39:22.700 align:middle line:84% So it's a metabolic syndrome, or otherwise called 00:39:22.700 --> 00:39:26.540 align:middle line:84% Syndrome X. Fasting glucose 133, that 00:39:26.540 --> 00:39:28.080 align:middle line:90% meets criteria for diabetes. 00:39:28.080 --> 00:39:32.270 align:middle line:84% Insulin is 27, which means this is high, but this is very high, 00:39:32.270 --> 00:39:35.180 align:middle line:84% bringing that glucose trying to compensate for that. 00:39:35.180 --> 00:39:37.130 align:middle line:90% Hemoglobin A1C is 7.3. 00:39:37.130 --> 00:39:39.980 align:middle line:90% The normal range is 4 to 6. 00:39:39.980 --> 00:39:42.320 align:middle line:84% And again to drive the whole point home, 00:39:42.320 --> 00:39:46.080 align:middle line:84% this is a 12-year-old girl with adult onset diabetes. 00:39:46.080 --> 00:39:49.130 align:middle line:90% So this is her body mass index. 00:39:49.130 --> 00:39:51.320 align:middle line:84% This is a 97th percentile curve, and this 00:39:51.320 --> 00:39:54.300 align:middle line:84% is where she is, well over the curve here. 00:39:54.300 --> 00:39:56.210 align:middle line:90% And now she has type 2 diabetes. 00:39:56.210 --> 00:39:58.370 align:middle line:84% She meets criteria for type 2 diabetes. 00:39:58.370 --> 00:40:00.080 align:middle line:84% She was not even entering seventh grade 00:40:00.080 --> 00:40:02.760 align:middle line:84% when I started doing summer vacation last year. 00:40:02.760 --> 00:40:06.410 align:middle line:84% 1 in 3 kids is going to be like this. 00:40:06.410 --> 00:40:07.380 align:middle line:90% That's a problem. 00:40:07.380 --> 00:40:11.990 align:middle line:84% So one way to study this problem is to study obesity, 00:40:11.990 --> 00:40:14.700 align:middle line:84% and one way to think about obesity related to diabetes. 00:40:14.700 --> 00:40:17.720 align:middle line:84% So one way to study obesity is to think about-- specifically 00:40:17.720 --> 00:40:21.140 align:middle line:84% what I'm interested in studying is how adipogenesis, 00:40:21.140 --> 00:40:22.910 align:middle line:84% or the process of making fat cells, 00:40:22.910 --> 00:40:26.270 align:middle line:84% might be related to how insulin works. 00:40:26.270 --> 00:40:28.400 align:middle line:84% And the best established way to study 00:40:28.400 --> 00:40:30.980 align:middle line:84% that is to look at the insulin receptor. 00:40:30.980 --> 00:40:33.570 align:middle line:84% So now we're zooming into one of these target tissues-- 00:40:33.570 --> 00:40:37.250 align:middle line:84% remember, liver, muscle, and, fat-- and the insulin receptor 00:40:37.250 --> 00:40:40.490 align:middle line:84% is a protein that sticks in the membrane of the cell. 00:40:40.490 --> 00:40:43.100 align:middle line:84% Like any other protein, it's got a gene coding for it 00:40:43.100 --> 00:40:46.520 align:middle line:84% in the DNA, gets transcribed, gets translated. 00:40:46.520 --> 00:40:49.400 align:middle line:84% The protein goes out to the surface. 00:40:49.400 --> 00:40:52.160 align:middle line:84% There are known mutations in the insulin receptor that 00:40:52.160 --> 00:40:54.860 align:middle line:84% can cause you to have problems with glucose handling 00:40:54.860 --> 00:40:55.590 align:middle line:90% and diabetes. 00:40:55.590 --> 00:40:57.007 align:middle line:84% In fact, there's a syndrome called 00:40:57.007 --> 00:41:00.080 align:middle line:84% leprechaunism, where these kids look like leprechauns actually. 00:41:00.080 --> 00:41:01.820 align:middle line:84% They die pretty quickly after birth, 00:41:01.820 --> 00:41:04.857 align:middle line:84% but they have mutations in the insulin receptor. 00:41:04.857 --> 00:41:07.190 align:middle line:84% It turns out you can actually be born without an insulin 00:41:07.190 --> 00:41:07.820 align:middle line:90% receptor. 00:41:07.820 --> 00:41:09.140 align:middle line:90% There's been a null person. 00:41:09.140 --> 00:41:10.910 align:middle line:84% There's been one patient, I think, 00:41:10.910 --> 00:41:13.310 align:middle line:84% that was found to have no insulin receptors. 00:41:13.310 --> 00:41:14.960 align:middle line:84% And obviously the insulin level was sky 00:41:14.960 --> 00:41:17.120 align:middle line:84% high in this baby at birth, because it's not 00:41:17.120 --> 00:41:18.852 align:middle line:90% acting anywhere. 00:41:18.852 --> 00:41:20.810 align:middle line:84% But what happens is-- what's supposed to happen 00:41:20.810 --> 00:41:22.910 align:middle line:84% is insulin binds to the insulin receptor, 00:41:22.910 --> 00:41:26.180 align:middle line:84% and then these phosphate groups are added to itself. 00:41:26.180 --> 00:41:29.330 align:middle line:84% And that changes the shape of this molecule in such a way 00:41:29.330 --> 00:41:31.880 align:middle line:84% that it can actually start to interact with other molecules. 00:41:31.880 --> 00:41:33.255 align:middle line:84% And in fact, the insulin receptor 00:41:33.255 --> 00:41:36.090 align:middle line:84% also adds phosphate groups to other things. 00:41:36.090 --> 00:41:39.320 align:middle line:84% In fact, these are insulin receptor substrate 1, 00:41:39.320 --> 00:41:41.660 align:middle line:84% insulin receptor substrate 2, 3, and 4, 00:41:41.660 --> 00:41:44.190 align:middle line:90% and there are some others here. 00:41:44.190 --> 00:41:46.820 align:middle line:84% So in collaboration with Ron Kahn, President of the Joslin 00:41:46.820 --> 00:41:48.650 align:middle line:84% Diabetes Center, he's created some mice 00:41:48.650 --> 00:41:51.350 align:middle line:84% where they've basically knocked out IRS1, 00:41:51.350 --> 00:41:53.670 align:middle line:90% 2, 3, and 4 in separate mice. 00:41:53.670 --> 00:41:56.090 align:middle line:84% So these otherwise normal mice missing this one 00:41:56.090 --> 00:41:57.870 align:middle line:90% particular gene. 00:41:57.870 --> 00:42:01.280 align:middle line:84% And if you take fat cells from these mice-- 00:42:01.280 --> 00:42:05.030 align:middle line:84% in fact, if you take pre-fat cells or pre-adipocytes, 00:42:05.030 --> 00:42:08.330 align:middle line:84% you can make them go into fat cells using a standard cocktail 00:42:08.330 --> 00:42:09.290 align:middle line:90% of hormones. 00:42:09.290 --> 00:42:11.870 align:middle line:84% This has been done for the past two decades. 00:42:11.870 --> 00:42:14.060 align:middle line:84% You take pre-fat cells and make them 00:42:14.060 --> 00:42:16.580 align:middle line:84% into fat cells with a particular cocktail. 00:42:16.580 --> 00:42:19.310 align:middle line:84% And what happens is if you take normal pre-fat cells 00:42:19.310 --> 00:42:22.910 align:middle line:84% and make them into fat cells and stain for the fat or the lipid, 00:42:22.910 --> 00:42:24.650 align:middle line:90% you get a big red circle here. 00:42:24.650 --> 00:42:26.750 align:middle line:90% This is a dish of the cells. 00:42:26.750 --> 00:42:29.870 align:middle line:84% If you knock out IRS 1 and try to do this, 00:42:29.870 --> 00:42:32.540 align:middle line:84% you cannot get the fat cells to form. 00:42:32.540 --> 00:42:35.330 align:middle line:84% So again, pre-fat cells, the same cocktail, 00:42:35.330 --> 00:42:36.920 align:middle line:90% you don't get fat cells in here. 00:42:36.920 --> 00:42:37.850 align:middle line:90% There's no stain. 00:42:37.850 --> 00:42:39.955 align:middle line:84% There's barely a stain for any red here. 00:42:39.955 --> 00:42:41.330 align:middle line:84% And it turns out if you knock out 00:42:41.330 --> 00:42:44.270 align:middle line:84% IRS3, 2, and 4 in the middle here, 00:42:44.270 --> 00:42:48.410 align:middle line:84% you get a gradation of the phenotype, interestingly. 00:42:48.410 --> 00:42:50.510 align:middle line:90% So it's not all or none. 00:42:50.510 --> 00:42:52.520 align:middle line:84% If you knock out the other IRS molecules, 00:42:52.520 --> 00:42:56.150 align:middle line:84% you get a gradation of the phenotype, curiously. 00:42:56.150 --> 00:42:59.000 align:middle line:84% So what we did is in collaboration with Ron Kahn, 00:42:59.000 --> 00:43:04.580 align:middle line:84% we actually made a list of genes that behaved in this pattern. 00:43:04.580 --> 00:43:06.612 align:middle line:84% We you don't have numbers here, but we're just 00:43:06.612 --> 00:43:08.820 align:middle line:84% trying to find genes that go along with that pattern. 00:43:08.820 --> 00:43:11.330 align:middle line:84% So let me show you an example of one of these things. 00:43:11.330 --> 00:43:12.140 align:middle line:90% Here's a gene. 00:43:12.140 --> 00:43:13.452 align:middle line:90% Here's wild type 1. 00:43:13.452 --> 00:43:14.160 align:middle line:90% Here's wild type. 00:43:14.160 --> 00:43:18.320 align:middle line:84% Here's IRS1 knockout, so no fat cells, easy to make fat cells. 00:43:18.320 --> 00:43:20.390 align:middle line:90% And we're taking the genes-- 00:43:20.390 --> 00:43:23.940 align:middle line:84% we're taking RNA before we try to make them into fat cells. 00:43:23.940 --> 00:43:27.650 align:middle line:84% The hypothesis here is, what genes are different 00:43:27.650 --> 00:43:30.560 align:middle line:84% before we try to make them into fat cells that 00:43:30.560 --> 00:43:36.170 align:middle line:84% might be impacting the process of making fat cells? 00:43:36.170 --> 00:43:38.212 align:middle line:84% Now we'll take a second here, and you're probably 00:43:38.212 --> 00:43:39.253 align:middle line:90% asking me, wait a minute. 00:43:39.253 --> 00:43:40.680 align:middle line:84% We were just talking about humans. 00:43:40.680 --> 00:43:42.805 align:middle line:84% We were just talking about this kid that was obese, 00:43:42.805 --> 00:43:44.480 align:middle line:84% and now we're talking about mice here. 00:43:44.480 --> 00:43:47.960 align:middle line:84% I'm going to bring this back to the humans in a minute, 00:43:47.960 --> 00:43:50.210 align:middle line:84% because I want to intersect this with other data sets 00:43:50.210 --> 00:43:51.920 align:middle line:90% and show you where I'm going. 00:43:51.920 --> 00:43:54.650 align:middle line:84% But suffice it to say we have about 80 genes that 00:43:54.650 --> 00:43:56.090 align:middle line:90% follow this kind of pattern. 00:43:56.090 --> 00:43:58.340 align:middle line:90% They're high in the normals. 00:43:58.340 --> 00:44:01.640 align:middle line:84% They're low in the IRS1 knockouts before adipogenesis, 00:44:01.640 --> 00:44:04.350 align:middle line:84% and they're in this kind of monotonic pattern here, 00:44:04.350 --> 00:44:06.980 align:middle line:90% or vice versa, going down or up. 00:44:06.980 --> 00:44:10.670 align:middle line:84% Came up with a list of 88 genes that fit this kind of pattern. 00:44:10.670 --> 00:44:14.070 align:middle line:84% Now put those 88 genes on hold for a second. 00:44:14.070 --> 00:44:16.035 align:middle line:84% Let's talk about a second disease. 00:44:16.035 --> 00:44:17.660 align:middle line:84% I happened to be involved with studying 00:44:17.660 --> 00:44:19.880 align:middle line:84% this particular syndrome called progeria. 00:44:19.880 --> 00:44:22.973 align:middle line:84% Progeria is advanced aging syndrome in children. 00:44:22.973 --> 00:44:24.890 align:middle line:84% In fact, it makes-- it's very rare, so about 1 00:44:24.890 --> 00:44:26.810 align:middle line:84% in 8,000,000 births, and it makes 00:44:26.810 --> 00:44:28.850 align:middle line:84% children look like they're getting older. 00:44:28.850 --> 00:44:32.427 align:middle line:84% We've all heard of these kids and this kind of story. 00:44:32.427 --> 00:44:34.010 align:middle line:84% And specifically they lose their hair. 00:44:34.010 --> 00:44:34.790 align:middle line:90% They have alopecia. 00:44:34.790 --> 00:44:35.790 align:middle line:90% They're certainly short. 00:44:35.790 --> 00:44:37.910 align:middle line:84% They're shorter than age-matched kids. 00:44:37.910 --> 00:44:40.460 align:middle line:84% That the skin changes that make them look old. 00:44:40.460 --> 00:44:41.910 align:middle line:90% They don't go through puberty. 00:44:41.910 --> 00:44:43.250 align:middle line:90% They have poor weight gain. 00:44:43.250 --> 00:44:45.500 align:middle line:84% And usually they die at age 15 because 00:44:45.500 --> 00:44:48.680 align:middle line:84% of cardiovascular problems, atherosclerosis, get strokes, 00:44:48.680 --> 00:44:50.030 align:middle line:90% they get heart attacks. 00:44:50.030 --> 00:44:52.640 align:middle line:84% Now even though they look older, they 00:44:52.640 --> 00:44:55.520 align:middle line:84% don't get everything associated with being old. 00:44:55.520 --> 00:44:56.960 align:middle line:90% They don't get Alzheimer's. 00:44:56.960 --> 00:44:58.790 align:middle line:90% They don't get cancer. 00:44:58.790 --> 00:45:01.400 align:middle line:84% It's only the atherosclerosis and these other changes 00:45:01.400 --> 00:45:04.040 align:middle line:90% here that they get. 00:45:04.040 --> 00:45:05.960 align:middle line:84% Now as you might know, the gene for this 00:45:05.960 --> 00:45:08.480 align:middle line:90% was just recently discovered. 00:45:08.480 --> 00:45:11.480 align:middle line:84% Very amazing that they found this gene. 00:45:11.480 --> 00:45:13.670 align:middle line:84% I got involved with this particular syndrome only 00:45:13.670 --> 00:45:15.560 align:middle line:84% because of my friend Leslie Gordon, 00:45:15.560 --> 00:45:17.500 align:middle line:84% who was a medical school classmate of mine 00:45:17.500 --> 00:45:19.650 align:middle line:90% and who has a son with this. 00:45:19.650 --> 00:45:20.690 align:middle line:90% And she's got an MD-PhD. 00:45:20.690 --> 00:45:22.640 align:middle line:84% Instead of just studying this disease herself, 00:45:22.640 --> 00:45:25.610 align:middle line:84% she built a foundation called the Progeria Research 00:45:25.610 --> 00:45:26.600 align:middle line:90% Foundation. 00:45:26.600 --> 00:45:28.880 align:middle line:84% Got Francis Collins interested, the guy 00:45:28.880 --> 00:45:32.330 align:middle line:84% who's basically doing the Genome Project, the head of NHGRI. 00:45:32.330 --> 00:45:34.640 align:middle line:84% And within 18 months of forming this foundation, 00:45:34.640 --> 00:45:36.770 align:middle line:90% they found the gene for this. 00:45:36.770 --> 00:45:39.380 align:middle line:90% Incredibly difficult problem. 00:45:39.380 --> 00:45:41.180 align:middle line:84% There are people who on the first day 00:45:41.180 --> 00:45:44.060 align:middle line:84% would not believe this is still a genetic syndrome. 00:45:44.060 --> 00:45:46.460 align:middle line:84% One in 8,000,000 births is way less 00:45:46.460 --> 00:45:49.820 align:middle line:90% than the mutation frequency. 00:45:49.820 --> 00:45:51.785 align:middle line:84% We have a whole bunch of diseases that at 1 00:45:51.785 --> 00:45:54.290 align:middle line:90% in 40,000, 1 in 100,000. 00:45:54.290 --> 00:45:57.530 align:middle line:84% 1 in 8,000,000 seems like it's so rare. 00:45:57.530 --> 00:46:00.000 align:middle line:84% How could a mutation be that rare? 00:46:00.000 --> 00:46:01.850 align:middle line:90% So it's very questionable. 00:46:01.850 --> 00:46:05.270 align:middle line:84% And secondly, they had a very hard time finding any families 00:46:05.270 --> 00:46:09.270 align:middle line:84% where two of these kids were in the same family. 00:46:09.270 --> 00:46:12.890 align:middle line:84% So it was dumb luck that they were able to find this gene, 00:46:12.890 --> 00:46:15.620 align:middle line:84% because in one of the individuals-- 00:46:15.620 --> 00:46:18.350 align:middle line:84% at any point there's only about 50 of these kids on the planet, 00:46:18.350 --> 00:46:19.850 align:middle line:90% really. 00:46:19.850 --> 00:46:22.820 align:middle line:84% And one of these kids, in one of the samples, someone 00:46:22.820 --> 00:46:25.400 align:middle line:84% with a sharp eye saw a piece of chromosome one 00:46:25.400 --> 00:46:26.900 align:middle line:90% was actually inverted. 00:46:26.900 --> 00:46:28.910 align:middle line:84% The bands, those banding patterns, 00:46:28.910 --> 00:46:31.383 align:middle line:84% was actually different in that one individual. 00:46:31.383 --> 00:46:33.800 align:middle line:84% And that got Francis Collins to think maybe it's something 00:46:33.800 --> 00:46:35.685 align:middle line:84% on that particular arm of chromosome 1, 00:46:35.685 --> 00:46:37.310 align:middle line:84% and they sequenced all the genes there, 00:46:37.310 --> 00:46:39.290 align:middle line:84% and they found the particular mutation. 00:46:39.290 --> 00:46:42.350 align:middle line:84% It's a gene called lamin A. And it's not 00:46:42.350 --> 00:46:43.670 align:middle line:90% just lamin A that causes it. 00:46:43.670 --> 00:46:45.110 align:middle line:84% Lamin A has actually been involved 00:46:45.110 --> 00:46:46.550 align:middle line:84% in a lot of other diseases, which 00:46:46.550 --> 00:46:47.758 align:middle line:90% we'll talk about in a second. 00:46:47.758 --> 00:46:51.680 align:middle line:84% It's just this one mutation in this one gene that causes it. 00:46:51.680 --> 00:46:54.870 align:middle line:84% Now amazingly enough, in the same issue of Nature, 00:46:54.870 --> 00:46:58.820 align:middle line:84% less than a year ago, the mouse model for this was published. 00:46:58.820 --> 00:47:00.410 align:middle line:90% How is that? 00:47:00.410 --> 00:47:04.430 align:middle line:84% Because this group at National Cancer Institute realized-- 00:47:04.430 --> 00:47:06.620 align:middle line:84% lamin A is known to be involved in certain forms 00:47:06.620 --> 00:47:07.970 align:middle line:90% of muscular dystrophy. 00:47:07.970 --> 00:47:09.920 align:middle line:84% They said, well, they wanted to create 00:47:09.920 --> 00:47:12.740 align:middle line:84% a mouse model of a particular form of muscular dystrophy. 00:47:12.740 --> 00:47:16.340 align:middle line:84% They accidentally created a mouse model for progeria. 00:47:16.340 --> 00:47:19.550 align:middle line:84% So the same issue-- you got the gene that causes it. 00:47:19.550 --> 00:47:21.410 align:middle line:90% You got diagnostic tests. 00:47:21.410 --> 00:47:25.910 align:middle line:84% And you got a mouse model to study it, and just in a tour de 00:47:25.910 --> 00:47:27.650 align:middle line:90% force issue there. 00:47:27.650 --> 00:47:29.720 align:middle line:84% Imagine how long it takes for these things 00:47:29.720 --> 00:47:31.520 align:middle line:90% to happen for a disease. 00:47:31.520 --> 00:47:34.590 align:middle line:84% Start to finish, 18 months here for something 00:47:34.590 --> 00:47:37.030 align:middle line:90% that affects maybe 50 people. 00:47:37.030 --> 00:47:39.530 align:middle line:84% So if you put your mind to it, you can solve these diseases. 00:47:39.530 --> 00:47:43.380 align:middle line:84% It's amazing what the tools and the people can do today. 00:47:43.380 --> 00:47:45.620 align:middle line:84% So in collaboration with the Progeria Foundation 00:47:45.620 --> 00:47:47.060 align:middle line:90% what we've done is this. 00:47:47.060 --> 00:47:48.650 align:middle line:84% It turns out over the past 20 years 00:47:48.650 --> 00:47:51.093 align:middle line:84% when people thought they had kids with progeria, that they 00:47:51.093 --> 00:47:52.760 align:middle line:84% were trying to treat kids with progeria, 00:47:52.760 --> 00:47:54.053 align:middle line:90% they had no idea what to do. 00:47:54.053 --> 00:47:55.970 align:middle line:84% So one thing they did is they took fibroblasts 00:47:55.970 --> 00:47:58.790 align:middle line:84% and they saved them in national repositories. 00:47:58.790 --> 00:48:01.850 align:middle line:84% So there's a cell repository called the Coriell cell 00:48:01.850 --> 00:48:04.960 align:middle line:84% repository where you can order fibroblast cell 00:48:04.960 --> 00:48:08.000 align:middle line:84% lines from patients with exceedingly rare disorders, 00:48:08.000 --> 00:48:09.610 align:middle line:90% including progeria. 00:48:09.610 --> 00:48:11.830 align:middle line:84% So we have gotten three cell lines 00:48:11.830 --> 00:48:15.130 align:middle line:84% from patients that were stored in the Coriell collection 00:48:15.130 --> 00:48:17.962 align:middle line:84% over the past 10 years, and we ordered them three times 00:48:17.962 --> 00:48:19.420 align:middle line:84% to make sure that we're controlling 00:48:19.420 --> 00:48:21.980 align:middle line:84% at least for passage number and things like that. 00:48:21.980 --> 00:48:24.820 align:middle line:84% And we compared them to age-matched normal fibroblasts 00:48:24.820 --> 00:48:26.620 align:middle line:84% from the same Coriell repository. 00:48:26.620 --> 00:48:28.870 align:middle line:84% Looked at what genes are different between progeria 00:48:28.870 --> 00:48:31.090 align:middle line:90% and non-progeria. 00:48:31.090 --> 00:48:33.220 align:middle line:84% And these are more up-to-date arrays. 00:48:33.220 --> 00:48:37.210 align:middle line:84% These are U-133 arrays on two-- the whole genome on two chips-- 00:48:37.210 --> 00:48:39.640 align:middle line:84% that gave us about 33,000 genes, and used 00:48:39.640 --> 00:48:41.570 align:middle line:84% a particular analytic technique, and we 00:48:41.570 --> 00:48:44.140 align:middle line:84% got about a list of about 366 genes 00:48:44.140 --> 00:48:46.480 align:middle line:84% that were different between fibroblasts 00:48:46.480 --> 00:48:49.630 align:middle line:84% in progeria and fibroblasts from age-matched controls. 00:48:49.630 --> 00:48:53.538 align:middle line:90% 00:48:53.538 --> 00:48:55.330 align:middle line:84% And so if you look at where these genes are 00:48:55.330 --> 00:48:57.220 align:middle line:84% and what categories they're involved with, 00:48:57.220 --> 00:48:59.860 align:middle line:84% you see interesting things like development, signal 00:48:59.860 --> 00:49:01.750 align:middle line:90% transduction, cell adhesion. 00:49:01.750 --> 00:49:03.950 align:middle line:84% The only thing I want to point out is this. 00:49:03.950 --> 00:49:06.970 align:middle line:84% These tables are made using a catalog called Gene Ontology. 00:49:06.970 --> 00:49:09.260 align:middle line:84% You're going to learn about that more in the future. 00:49:09.260 --> 00:49:11.290 align:middle line:84% And every week or every other week, 00:49:11.290 --> 00:49:14.020 align:middle line:84% the list of known properties of genes and proteins 00:49:14.020 --> 00:49:15.470 align:middle line:90% gets updated. 00:49:15.470 --> 00:49:20.830 align:middle line:84% But still the majority of genes unclassified, 161. 00:49:20.830 --> 00:49:23.200 align:middle line:84% No matter how interesting or how comprehensive 00:49:23.200 --> 00:49:26.200 align:middle line:84% we think Gene Ontology is, the vast majority of genes 00:49:26.200 --> 00:49:27.400 align:middle line:90% just have no-- 00:49:27.400 --> 00:49:29.560 align:middle line:84% we have no idea where they are, what 00:49:29.560 --> 00:49:32.680 align:middle line:90% they do, what roles they play. 00:49:32.680 --> 00:49:34.420 align:middle line:84% These are where the proteins that they're 00:49:34.420 --> 00:49:36.700 align:middle line:90% coding for actually take place. 00:49:36.700 --> 00:49:41.020 align:middle line:84% Nucleus, membrane, unclassified 131 again. 00:49:41.020 --> 00:49:43.750 align:middle line:84% No idea where these proteins even are in the cell. 00:49:43.750 --> 00:49:45.340 align:middle line:90% That's life today. 00:49:45.340 --> 00:49:48.220 align:middle line:84% We have a great taxonomy here of all of these terms 00:49:48.220 --> 00:49:50.680 align:middle line:84% and structure vocabulary, but we have no-- there's 00:49:50.680 --> 00:49:55.160 align:middle line:84% very little data for most genes in this still. 00:49:55.160 --> 00:49:56.750 align:middle line:84% And there's a bunch of genes that 00:49:56.750 --> 00:50:00.170 align:middle line:84% were upregulated that seemed to already be known to play 00:50:00.170 --> 00:50:02.480 align:middle line:90% a role in atherosclerosis. 00:50:02.480 --> 00:50:04.790 align:middle line:84% So this paper, again, talking about genome medicine, 00:50:04.790 --> 00:50:06.800 align:middle line:84% this paper is going to end with this list, 00:50:06.800 --> 00:50:08.828 align:middle line:84% only because a number of these are targets, 00:50:08.828 --> 00:50:10.370 align:middle line:84% and pharmaceutical companies actually 00:50:10.370 --> 00:50:13.760 align:middle line:84% have drugs against a number of these already. 00:50:13.760 --> 00:50:16.280 align:middle line:84% For kids with progeria it's basically a death sentence 00:50:16.280 --> 00:50:16.780 align:middle line:90% today. 00:50:16.780 --> 00:50:18.110 align:middle line:90% We have no medicines. 00:50:18.110 --> 00:50:20.330 align:middle line:84% We treat them with nothing today. 00:50:20.330 --> 00:50:22.833 align:middle line:84% Maybe some people treat one or two kids with growth hormone 00:50:22.833 --> 00:50:24.500 align:middle line:84% to see if growth hormone does something, 00:50:24.500 --> 00:50:26.917 align:middle line:84% but that's basically like a voodoo medicine kind of thing. 00:50:26.917 --> 00:50:29.000 align:middle line:84% We have no proof of that doing anything. 00:50:29.000 --> 00:50:31.235 align:middle line:84% Now we can at least get to some information 00:50:31.235 --> 00:50:34.880 align:middle line:84% to possibly even think about clinical trials here. 00:50:34.880 --> 00:50:36.920 align:middle line:90% But let's be frank here. 00:50:36.920 --> 00:50:40.700 align:middle line:84% These were fibroblasts that were frozen for years in the cell 00:50:40.700 --> 00:50:41.925 align:middle line:90% repository. 00:50:41.925 --> 00:50:43.550 align:middle line:84% And now we're making a claim as to what 00:50:43.550 --> 00:50:45.203 align:middle line:90% their blood vessels look like. 00:50:45.203 --> 00:50:47.120 align:middle line:84% And not just the blood vessels, but the intima 00:50:47.120 --> 00:50:48.710 align:middle line:90% of their blood vessels. 00:50:48.710 --> 00:50:50.060 align:middle line:90% This is not the same cell. 00:50:50.060 --> 00:50:53.330 align:middle line:84% The fibroblasts might have been taken from their cheek, 00:50:53.330 --> 00:50:55.490 align:middle line:84% but we're just making a claim here because we don't 00:50:55.490 --> 00:50:56.870 align:middle line:90% have access to the real tissue. 00:50:56.870 --> 00:50:59.920 align:middle line:90% 00:50:59.920 --> 00:51:02.580 align:middle line:84% So one of the reasons why I'm mentioning progeria 00:51:02.580 --> 00:51:05.280 align:middle line:84% here is remember that in picture you just saw, 00:51:05.280 --> 00:51:07.530 align:middle line:84% one of the ways that these kids get diagnosed 00:51:07.530 --> 00:51:09.930 align:middle line:84% is because by the time they get to age two, 00:51:09.930 --> 00:51:11.500 align:middle line:90% they start to lose their hair. 00:51:11.500 --> 00:51:16.390 align:middle line:84% But even more interestingly is they lose all their fat cells. 00:51:16.390 --> 00:51:20.640 align:middle line:84% It turns out in progeria you lose all your subcutaneous fat. 00:51:20.640 --> 00:51:23.160 align:middle line:84% And by death at autopsy it's very tough 00:51:23.160 --> 00:51:26.370 align:middle line:90% to find any fat in the body. 00:51:26.370 --> 00:51:28.140 align:middle line:84% It makes you think something is up 00:51:28.140 --> 00:51:32.320 align:middle line:84% with the actual process of making fat cells in these kids. 00:51:32.320 --> 00:51:33.840 align:middle line:84% In one study these kids have been 00:51:33.840 --> 00:51:35.430 align:middle line:84% known to have some insulin resistance, 00:51:35.430 --> 00:51:37.960 align:middle line:84% but it's just one study almost 10 years ago. 00:51:37.960 --> 00:51:40.770 align:middle line:84% In the mouse model, the last paragraph of that Nature paper 00:51:40.770 --> 00:51:43.860 align:middle line:84% says that they saw some process where muscle was 00:51:43.860 --> 00:51:45.168 align:middle line:90% re-differentiating into fat. 00:51:45.168 --> 00:51:47.460 align:middle line:84% They said they think there's some developmental problem 00:51:47.460 --> 00:51:49.440 align:middle line:84% going on, but it's not very clear. 00:51:49.440 --> 00:51:52.540 align:middle line:84% Other mutations in lamin A cause lipodystrophy, 00:51:52.540 --> 00:51:54.952 align:middle line:84% which is a syndrome where you lose all your fat cells. 00:51:54.952 --> 00:51:56.910 align:middle line:84% So it makes us think that there is legitimately 00:51:56.910 --> 00:51:59.160 align:middle line:84% something wrong with fat cell development in kids 00:51:59.160 --> 00:52:01.230 align:middle line:90% with progeria. 00:52:01.230 --> 00:52:04.080 align:middle line:90% So now I told you about-- 00:52:04.080 --> 00:52:04.830 align:middle line:90% where are we here? 00:52:04.830 --> 00:52:05.920 align:middle line:90% Hold on. 00:52:05.920 --> 00:52:09.240 align:middle line:84% I told you about 80 or 90 genes that were in the mouse models 00:52:09.240 --> 00:52:13.260 align:middle line:84% going along with fat, so about 360 genes that 00:52:13.260 --> 00:52:16.410 align:middle line:84% are different between progeria kids and age-matched controls, 00:52:16.410 --> 00:52:18.900 align:middle line:84% and they have some problem with fat. 00:52:18.900 --> 00:52:20.520 align:middle line:84% I have a third data set here, which 00:52:20.520 --> 00:52:22.110 align:middle line:84% was published by Gary Ruvkun's group 00:52:22.110 --> 00:52:23.670 align:middle line:90% over at Mass General Hospital. 00:52:23.670 --> 00:52:25.470 align:middle line:90% This came out about last year. 00:52:25.470 --> 00:52:27.810 align:middle line:84% And now this is in a totally different species. 00:52:27.810 --> 00:52:29.310 align:middle line:90% This is in the worm. 00:52:29.310 --> 00:52:32.160 align:middle line:84% To give you an idea of what we can do today, 00:52:32.160 --> 00:52:35.790 align:middle line:84% what this group did is they knocked out every single gene 00:52:35.790 --> 00:52:37.960 align:middle line:90% in the worm genome. 00:52:37.960 --> 00:52:41.550 align:middle line:84% So a worm has about 16,000, 17,000 genes, and they 00:52:41.550 --> 00:52:43.350 align:middle line:84% basically serially knocked out each 00:52:43.350 --> 00:52:45.180 align:middle line:84% and every one of these genes to look 00:52:45.180 --> 00:52:47.100 align:middle line:90% at what happens to the worm. 00:52:47.100 --> 00:52:49.330 align:middle line:84% Now in a worm it's actually pretty easy to do. 00:52:49.330 --> 00:52:52.710 align:middle line:84% You can actually get bacteria to make a particular sequence 00:52:52.710 --> 00:52:56.550 align:middle line:84% called an RNAi that interferes with the worm's ability 00:52:56.550 --> 00:52:57.782 align:middle line:90% to make that gene. 00:52:57.782 --> 00:52:59.490 align:middle line:84% So you don't even have to change the DNA. 00:52:59.490 --> 00:53:01.350 align:middle line:84% You could just effectively shut down 00:53:01.350 --> 00:53:04.170 align:middle line:84% the amount of that gene that's being made in the worm. 00:53:04.170 --> 00:53:05.490 align:middle line:90% And basically it's very easy. 00:53:05.490 --> 00:53:07.240 align:middle line:84% You don't even have to inject these worms. 00:53:07.240 --> 00:53:11.610 align:middle line:84% You essentially set up 16,700 buckets of bacteria, 00:53:11.610 --> 00:53:14.250 align:middle line:84% and just put the worms in it, and they eat the bacteria, 00:53:14.250 --> 00:53:17.190 align:middle line:84% and basically they knock out the gene themselves. 00:53:17.190 --> 00:53:18.893 align:middle line:90% That's literally how it works. 00:53:18.893 --> 00:53:20.310 align:middle line:84% All you have to do is get the worm 00:53:20.310 --> 00:53:22.445 align:middle line:84% to eat the bacteria that makes the RNAi. 00:53:22.445 --> 00:53:23.070 align:middle line:90% It's very easy. 00:53:23.070 --> 00:53:25.717 align:middle line:84% It knocks out the gene in that worm. 00:53:25.717 --> 00:53:28.050 align:middle line:84% So what these guys did is-- they had back-to-back Nature 00:53:28.050 --> 00:53:29.222 align:middle line:90% papers on this process. 00:53:29.222 --> 00:53:31.680 align:middle line:84% The first one was on a whole bunch of different phenotypes, 00:53:31.680 --> 00:53:35.170 align:middle line:84% but the second paper was looking at that, amazingly enough. 00:53:35.170 --> 00:53:37.110 align:middle line:84% And here there's a picture of a worm, 00:53:37.110 --> 00:53:39.840 align:middle line:84% and they're using the same stain for lipid 00:53:39.840 --> 00:53:42.290 align:middle line:84% that I showed you two or three slides ago, what 00:53:42.290 --> 00:53:43.415 align:middle line:90% we were using on the mouse. 00:53:43.415 --> 00:53:45.165 align:middle line:84% And basically they're saying, for example, 00:53:45.165 --> 00:53:47.685 align:middle line:84% if you knock out daf-2, you get an increase in the fat. 00:53:47.685 --> 00:53:49.770 align:middle line:84% If you knock out daf-2 and daf-16, 00:53:49.770 --> 00:53:53.490 align:middle line:84% you get a subsequent decrease, daf-2 and daf-3, slightly more. 00:53:53.490 --> 00:53:55.980 align:middle line:84% And they can quantitate the amount of fat 00:53:55.980 --> 00:53:59.550 align:middle line:84% that results after you've knocked out any of these genes. 00:53:59.550 --> 00:54:01.050 align:middle line:84% And essentially they've made a list. 00:54:01.050 --> 00:54:03.430 align:middle line:84% It's available on the internet, amazingly. 00:54:03.430 --> 00:54:05.190 align:middle line:84% If you knock out any of these 112 genes, 00:54:05.190 --> 00:54:07.483 align:middle line:84% it increases the amount of fat in the worm. 00:54:07.483 --> 00:54:09.150 align:middle line:84% If you knock out any of these 305 genes, 00:54:09.150 --> 00:54:12.120 align:middle line:84% it decreases the amount of fat in the worm. 00:54:12.120 --> 00:54:16.140 align:middle line:84% So now we have three big data sets, one in the mouse, 00:54:16.140 --> 00:54:18.900 align:middle line:84% one in the human, one in the worm, 00:54:18.900 --> 00:54:22.607 align:middle line:84% two microarrays, one RNAi, spanning all of these. 00:54:22.607 --> 00:54:24.690 align:middle line:84% But they all have something to do with each other, 00:54:24.690 --> 00:54:29.670 align:middle line:84% because they're all thinking about fat and fat storage. 00:54:29.670 --> 00:54:32.460 align:middle line:84% Fat storage in the worm is very different than a human, 00:54:32.460 --> 00:54:35.292 align:middle line:84% but still a number of the genes are actually intersecting. 00:54:35.292 --> 00:54:36.750 align:middle line:84% And really what we're talking about 00:54:36.750 --> 00:54:40.800 align:middle line:84% is intersecting these three data sets. 00:54:40.800 --> 00:54:43.650 align:middle line:84% We got the worm, we got the mouse, we get the progeria, 00:54:43.650 --> 00:54:45.840 align:middle line:84% and we're looking at the intersection of all these 00:54:45.840 --> 00:54:46.890 align:middle line:90% three. 00:54:46.890 --> 00:54:49.140 align:middle line:84% So this is what I've been working on over the past six 00:54:49.140 --> 00:54:50.070 align:middle line:90% months, actually. 00:54:50.070 --> 00:54:52.380 align:middle line:84% And this is what we're calling integrative biology 00:54:52.380 --> 00:54:55.350 align:middle line:90% or integrative genomics. 00:54:55.350 --> 00:54:57.990 align:middle line:84% Because these large data sets are not-- 00:54:57.990 --> 00:55:01.170 align:middle line:84% they don't just exist, and a large majority they're actually 00:55:01.170 --> 00:55:02.352 align:middle line:90% publicly available. 00:55:02.352 --> 00:55:04.560 align:middle line:84% You could go to the National Center for Biotechnology 00:55:04.560 --> 00:55:08.190 align:middle line:84% Information, the NCBI GEO, and pull down any number 00:55:08.190 --> 00:55:11.037 align:middle line:90% of these kinds of data sets. 00:55:11.037 --> 00:55:12.120 align:middle line:90% And you could do it today. 00:55:12.120 --> 00:55:13.287 align:middle line:90% It's for your final project. 00:55:13.287 --> 00:55:16.380 align:middle line:84% You can go download any of 600 experiments 00:55:16.380 --> 00:55:18.870 align:middle line:84% where the microarrays have already been collected. 00:55:18.870 --> 00:55:23.490 align:middle line:84% The only hard part now is to have the interesting question. 00:55:23.490 --> 00:55:26.010 align:middle line:84% You can't just pair something with something else. 00:55:26.010 --> 00:55:27.660 align:middle line:84% You want to have to actually think 00:55:27.660 --> 00:55:30.342 align:middle line:84% about an interesting question, because the resources 00:55:30.342 --> 00:55:31.800 align:middle line:84% in all likelihood are already there 00:55:31.800 --> 00:55:33.930 align:middle line:84% to actually try to answer your question. 00:55:33.930 --> 00:55:35.970 align:middle line:84% Whether you're interested in circadian rhythms, 00:55:35.970 --> 00:55:39.090 align:middle line:84% or fat development, or anything, that's the hard part now, 00:55:39.090 --> 00:55:41.520 align:middle line:84% is knowing enough biology to actually ask 00:55:41.520 --> 00:55:43.050 align:middle line:84% an interesting question in this way. 00:55:43.050 --> 00:55:45.245 align:middle line:90% 00:55:45.245 --> 00:55:47.620 align:middle line:84% So it turns out if you look at the genes in intersection, 00:55:47.620 --> 00:55:50.620 align:middle line:84% just to give you an idea, these WNT genes-- 00:55:50.620 --> 00:55:53.530 align:middle line:84% WNT is actually a hormone that's made. 00:55:53.530 --> 00:55:55.030 align:middle line:84% WNT genes are actually well known 00:55:55.030 --> 00:55:56.140 align:middle line:84% to be involved with adipogenesis, 00:55:56.140 --> 00:55:58.140 align:middle line:84% and they are actually living in the intersection 00:55:58.140 --> 00:55:59.150 align:middle line:90% of these lists. 00:55:59.150 --> 00:56:02.590 align:middle line:84% So the worm egl2, which is called WNT7B, 00:56:02.590 --> 00:56:06.010 align:middle line:84% increases the amount of fat if it's knocked out. 00:56:06.010 --> 00:56:09.460 align:middle line:84% The mouse WNT6 and 10A increase with dysadipogenesis, 00:56:09.460 --> 00:56:11.230 align:middle line:90% depending on the IRS knockouts. 00:56:11.230 --> 00:56:14.598 align:middle line:84% And WNT5A and WNT7B are down in progeria. 00:56:14.598 --> 00:56:17.140 align:middle line:84% So that's an example of a gene that lives in the intersection 00:56:17.140 --> 00:56:18.618 align:middle line:90% of these three data sets. 00:56:18.618 --> 00:56:21.160 align:middle line:84% I'm just amazed that we got any of these things to intersect, 00:56:21.160 --> 00:56:22.368 align:middle line:90% because they're so different. 00:56:22.368 --> 00:56:23.680 align:middle line:90% Three different species. 00:56:23.680 --> 00:56:26.140 align:middle line:84% We're not even looking at the fat cells here. 00:56:26.140 --> 00:56:27.640 align:middle line:84% With the progeria kids we're looking 00:56:27.640 --> 00:56:29.530 align:middle line:84% at a fibroblast, which is a proxy 00:56:29.530 --> 00:56:33.850 align:middle line:84% for the pre-adipocyte, which is a proxy for the adipocyte. 00:56:33.850 --> 00:56:36.370 align:middle line:84% But still you can get some signal this way. 00:56:36.370 --> 00:56:40.140 align:middle line:90% 00:56:40.140 --> 00:56:41.390 align:middle line:90% This picture is just the same. 00:56:41.390 --> 00:56:43.098 align:middle line:84% Even my own daughter knows the difference 00:56:43.098 --> 00:56:45.650 align:middle line:84% between a human, a mouse, and a worm. 00:56:45.650 --> 00:56:47.102 align:middle line:84% So there's caveats to all of this. 00:56:47.102 --> 00:56:49.310 align:middle line:84% So we're doing a lot of this validation ourselves now 00:56:49.310 --> 00:56:51.740 align:middle line:84% in the bioinformatics group over at Children's. 00:56:51.740 --> 00:56:54.320 align:middle line:84% But the point here is integrative biology. 00:56:54.320 --> 00:56:56.120 align:middle line:84% To me this is where I think a lot of fun 00:56:56.120 --> 00:56:59.270 align:middle line:84% is going to happen for me in the next few years. 00:56:59.270 --> 00:57:00.800 align:middle line:84% We have a lot of these data sets, 00:57:00.800 --> 00:57:02.820 align:middle line:84% and now I'm trying to come up with interesting questions 00:57:02.820 --> 00:57:04.280 align:middle line:84% so I can put these together to try 00:57:04.280 --> 00:57:07.708 align:middle line:84% to get a massive handle on these processes 00:57:07.708 --> 00:57:10.250 align:middle line:84% that other people are just using one of these days that might 00:57:10.250 --> 00:57:11.987 align:middle line:90% not be able to get a grasp on. 00:57:11.987 --> 00:57:13.070 align:middle line:90% We've got gene expression. 00:57:13.070 --> 00:57:14.190 align:middle line:90% We've got proteomic data. 00:57:14.190 --> 00:57:15.190 align:middle line:90% We've talked about that. 00:57:15.190 --> 00:57:17.008 align:middle line:90% We've had genome scans. 00:57:17.008 --> 00:57:19.550 align:middle line:84% Joel Hirschhorn is going to be talking about how you actually 00:57:19.550 --> 00:57:21.092 align:middle line:84% try to find genes that are associated 00:57:21.092 --> 00:57:23.780 align:middle line:84% with a particular phenotype using SNPs. 00:57:23.780 --> 00:57:25.620 align:middle line:84% We have the clinical measurements. 00:57:25.620 --> 00:57:27.890 align:middle line:84% We have all of these data points that are collected 00:57:27.890 --> 00:57:30.890 align:middle line:90% in all of these hospitals. 00:57:30.890 --> 00:57:32.855 align:middle line:84% I'll talk about each of these separately. 00:57:32.855 --> 00:57:35.330 align:middle line:84% ENU mutagenesis is a new technique where-- 00:57:35.330 --> 00:57:36.680 align:middle line:90% it's actually not so new-- 00:57:36.680 --> 00:57:40.640 align:middle line:84% but you could take mice and give them this particular chemical, 00:57:40.640 --> 00:57:45.360 align:middle line:84% ethyl-nitrosourea, that causes mutations in their gametes. 00:57:45.360 --> 00:57:48.680 align:middle line:84% So then all of their offspring basically are mutants. 00:57:48.680 --> 00:57:51.140 align:middle line:84% And you do this for a couple of generations, 00:57:51.140 --> 00:57:54.410 align:middle line:84% and you find the interesting phenotype that-- 00:57:54.410 --> 00:57:56.360 align:middle line:84% wow, all of a sudden, this one has diabetes. 00:57:56.360 --> 00:57:59.720 align:middle line:84% You go back and see, what did we mutate? 00:57:59.720 --> 00:58:02.480 align:middle line:84% We basically mutate it to see as much as you can. 00:58:02.480 --> 00:58:04.760 align:middle line:84% You characterize the output and then go back 00:58:04.760 --> 00:58:07.610 align:middle line:84% to see what-- exactly where did we put the mutation in? 00:58:07.610 --> 00:58:09.440 align:middle line:90% That's ENU mutagenesis. 00:58:09.440 --> 00:58:12.260 align:middle line:84% RNAi we start to talk about with the worm, where you basically 00:58:12.260 --> 00:58:15.170 align:middle line:84% can just knock out any particular gene very quickly. 00:58:15.170 --> 00:58:18.770 align:middle line:84% People are starting to do this with humans and specifically 00:58:18.770 --> 00:58:20.690 align:middle line:84% with mouse models, but there's already 00:58:20.690 --> 00:58:22.310 align:middle line:84% a number of pharmaceutical companies 00:58:22.310 --> 00:58:24.470 align:middle line:84% within a five-mile radius of here trying to come up 00:58:24.470 --> 00:58:25.790 align:middle line:90% with drugs like that. 00:58:25.790 --> 00:58:28.768 align:middle line:84% And of course, most importantly, the prior biological knowledge. 00:58:28.768 --> 00:58:31.310 align:middle line:84% This is what's going to drive the interesting questions here. 00:58:31.310 --> 00:58:34.100 align:middle line:90% 00:58:34.100 --> 00:58:35.560 align:middle line:90% Any questions about that so far? 00:58:35.560 --> 00:58:41.140 align:middle line:90% 00:58:41.140 --> 00:58:43.385 align:middle line:90% All right. 00:58:43.385 --> 00:58:45.340 align:middle line:84% I'm going to talk about something just 00:58:45.340 --> 00:58:49.570 align:middle line:84% to bring this home so we can talk about-- 00:58:49.570 --> 00:58:53.020 align:middle line:84% at least you can see how easy it is to actually do this. 00:58:53.020 --> 00:58:54.730 align:middle line:84% Let me skip a couple of slides here. 00:58:54.730 --> 00:59:00.450 align:middle line:90% 00:59:00.450 --> 00:59:02.633 align:middle line:84% So yeah, we can actually talk about some of this. 00:59:02.633 --> 00:59:04.800 align:middle line:84% So let's say you're interested in doing some of this 00:59:04.800 --> 00:59:07.230 align:middle line:84% for a final project, that you want to actually put some 00:59:07.230 --> 00:59:09.012 align:middle line:90% of these data sets together. 00:59:09.012 --> 00:59:10.470 align:middle line:84% There are quite a number of sources 00:59:10.470 --> 00:59:14.130 align:middle line:84% where you can get microarray data and other genome size data 00:59:14.130 --> 00:59:16.930 align:middle line:90% today. 00:59:16.930 --> 00:59:19.090 align:middle line:84% And if you see some things in the description 00:59:19.090 --> 00:59:21.090 align:middle line:84% that you're actually biologically interested in, 00:59:21.090 --> 00:59:24.070 align:middle line:84% you might be able to run with that for a final project. 00:59:24.070 --> 00:59:26.730 align:middle line:84% I just want to point out a few of these data sources 00:59:26.730 --> 00:59:29.640 align:middle line:84% and then talk about one way how you can put this together 00:59:29.640 --> 00:59:32.340 align:middle line:90% even just using a web browser. 00:59:32.340 --> 00:59:34.350 align:middle line:84% So the cardiogenomics program is a source 00:59:34.350 --> 00:59:38.340 align:middle line:84% where I think there's over 150 microarrays now. 00:59:38.340 --> 00:59:40.950 align:middle line:84% The headquarters of this is over at Beth Israel Deaconess 00:59:40.950 --> 00:59:41.970 align:middle line:90% across the street. 00:59:41.970 --> 00:59:44.130 align:middle line:84% They have mouse models of cardiac development, 00:59:44.130 --> 00:59:46.470 align:middle line:84% so basically they have embryos, and they 00:59:46.470 --> 00:59:47.760 align:middle line:90% have sections of their heart. 00:59:47.760 --> 00:59:49.802 align:middle line:84% And basically all the genes are turned on and off 00:59:49.802 --> 00:59:51.780 align:middle line:84% at different parts during development. 00:59:51.780 --> 00:59:54.520 align:middle line:84% They have all sorts of different components of cardiac growth. 00:59:54.520 --> 00:59:56.478 align:middle line:84% So for example, people know that if one pathway 00:59:56.478 --> 00:59:58.562 align:middle line:84% or another pathway is involved in a heart failure, 00:59:58.562 --> 01:00:00.960 align:middle line:84% they have mouse models where those components are already 01:00:00.960 --> 01:00:02.160 align:middle line:90% knocked out. 01:00:02.160 --> 01:00:04.720 align:middle line:84% And they have a bunch of measures made in time series. 01:00:04.720 --> 01:00:07.440 align:middle line:84% So they take a normal mice, and they put them in this tank 01:00:07.440 --> 01:00:09.390 align:middle line:84% to swim, and they make them practice 01:00:09.390 --> 01:00:10.830 align:middle line:84% swimming a certain number of hours 01:00:10.830 --> 01:00:12.750 align:middle line:84% for weeks then look at their hearts. 01:00:12.750 --> 01:00:15.120 align:middle line:84% And this is what swimming does your heart, for example. 01:00:15.120 --> 01:00:18.810 align:middle line:84% They have those lists of genes available today. 01:00:18.810 --> 01:00:20.310 align:middle line:84% These slides are going to be online, 01:00:20.310 --> 01:00:22.693 align:middle line:84% so you don't have to copy down every single URL. 01:00:22.693 --> 01:00:24.360 align:middle line:84% If you just even write down the heading, 01:00:24.360 --> 01:00:26.430 align:middle line:90% then which one to look for. 01:00:26.430 --> 01:00:28.425 align:middle line:84% The Whitehead Institute, obviously the mother 01:00:28.425 --> 01:00:30.060 align:middle line:90% lode of these arrays. 01:00:30.060 --> 01:00:33.360 align:middle line:84% At least 12, maybe 14 publications now. 01:00:33.360 --> 01:00:35.220 align:middle line:84% Many different types of cancer, that's 01:00:35.220 --> 01:00:36.820 align:middle line:90% their primary focus there. 01:00:36.820 --> 01:00:38.850 align:middle line:84% A lot of clinical measurements with some. 01:00:38.850 --> 01:00:40.687 align:middle line:84% A lot of these are clinical samples, 01:00:40.687 --> 01:00:42.270 align:middle line:84% but they might have one or two columns 01:00:42.270 --> 01:00:43.380 align:middle line:90% of clinical measurements. 01:00:43.380 --> 01:00:44.430 align:middle line:90% Was this a smoker? 01:00:44.430 --> 01:00:46.320 align:middle line:90% Was this a non-smoker? 01:00:46.320 --> 01:00:47.190 align:middle line:90% That kind of thing. 01:00:47.190 --> 01:00:49.980 align:middle line:84% How long did this patient relapse, not relapse? 01:00:49.980 --> 01:00:54.330 align:middle line:84% But you've got leukemias, you've got solid tumors here, 01:00:54.330 --> 01:00:55.450 align:middle line:90% brain cancers. 01:00:55.450 --> 01:00:59.525 align:middle line:84% Things like that are in this data set. 01:00:59.525 --> 01:01:01.400 align:middle line:84% DC Children's, or Children's National Medical 01:01:01.400 --> 01:01:06.140 align:middle line:84% Center, 500 arrays for many interesting human diseases. 01:01:06.140 --> 01:01:08.630 align:middle line:84% Muscular dystrophy, dermatomyositis, so that's 01:01:08.630 --> 01:01:09.307 align:middle line:90% rare. 01:01:09.307 --> 01:01:10.640 align:middle line:90% A bunch of rheumatologic things. 01:01:10.640 --> 01:01:14.083 align:middle line:84% Heart failure as well as mouse, rat, and dog models 01:01:14.083 --> 01:01:15.500 align:middle line:84% of spinal cord injury, if you want 01:01:15.500 --> 01:01:17.870 align:middle line:84% to see what are the similarities and differences there. 01:01:17.870 --> 01:01:21.510 align:middle line:84% Pulmonary disease, including asthma and heart failure. 01:01:21.510 --> 01:01:24.870 align:middle line:84% I mentioned heart failure twice there. 01:01:24.870 --> 01:01:27.590 align:middle line:84% So that's DC Children's and Johns Hopkins. 01:01:27.590 --> 01:01:31.940 align:middle line:84% The Human Gene Expression Index I think is Brigham and Women's. 01:01:31.940 --> 01:01:34.640 align:middle line:84% 121 microarrays just from normal tissues, 01:01:34.640 --> 01:01:36.980 align:middle line:90% 19 normal human tissues. 01:01:36.980 --> 01:01:40.910 align:middle line:84% It's also an interesting list to think about. 01:01:40.910 --> 01:01:42.560 align:middle line:84% Stanford's microarrays is probably 01:01:42.560 --> 01:01:44.957 align:middle line:84% the largest besides the GEO, which 01:01:44.957 --> 01:01:46.040 align:middle line:90% we'll talk about a second. 01:01:46.040 --> 01:01:48.420 align:middle line:84% Stanford has probably close to 4,000 arrays 01:01:48.420 --> 01:01:50.810 align:middle line:84% now, measured across 11 species, covering 01:01:50.810 --> 01:01:53.300 align:middle line:90% more than 80 publications. 01:01:53.300 --> 01:01:55.880 align:middle line:84% Many different fields, most of which are yeast, 01:01:55.880 --> 01:01:57.920 align:middle line:84% but a whole bunch of human ones as well. 01:01:57.920 --> 01:02:01.430 align:middle line:84% The National NCBI GEO, it's basically right next 01:02:01.430 --> 01:02:02.300 align:middle line:90% to PubMed now. 01:02:02.300 --> 01:02:04.040 align:middle line:90% It's the same website. 01:02:04.040 --> 01:02:07.610 align:middle line:84% Over 8,000 arrays from over 100 different types of microarrays. 01:02:07.610 --> 01:02:09.320 align:middle line:90% Never mind how many experiments. 01:02:09.320 --> 01:02:11.730 align:middle line:84% Their database can handle over 100 different companies, 01:02:11.730 --> 01:02:14.750 align:middle line:84% 100 different special microarray products that can 01:02:14.750 --> 01:02:16.940 align:middle line:90% integrate across all of these. 01:02:16.940 --> 01:02:22.370 align:middle line:84% TREX is from TiGER, which is The Institute for Genome Research 01:02:22.370 --> 01:02:25.130 align:middle line:84% over in Gaithersburg in Maryland, outside of NIH. 01:02:25.130 --> 01:02:28.100 align:middle line:84% 500, 600 arrays for mouse and rat 01:02:28.100 --> 01:02:30.800 align:middle line:84% models of sleep, infection, hypertension, pulmonary 01:02:30.800 --> 01:02:32.155 align:middle line:90% disease. 01:02:32.155 --> 01:02:34.280 align:middle line:84% So like for example, if you're interested in asthma 01:02:34.280 --> 01:02:36.447 align:middle line:84% and you want to see differences between human asthma 01:02:36.447 --> 01:02:39.470 align:middle line:84% and the mouse model, the data is there today. 01:02:39.470 --> 01:02:43.550 align:middle line:84% I will guarantee you no one else has done that intersection. 01:02:43.550 --> 01:02:45.890 align:middle line:84% If you find some interesting genes 01:02:45.890 --> 01:02:47.960 align:middle line:84% that you think might code for proteins, that 01:02:47.960 --> 01:02:49.880 align:middle line:84% might be in the blood and you can use that as a measurement, 01:02:49.880 --> 01:02:52.295 align:middle line:84% you can get some interesting person, one of any collaborator 01:02:52.295 --> 01:02:53.990 align:middle line:84% here in the Longwood area to actually measure 01:02:53.990 --> 01:02:55.290 align:middle line:90% this in the blood, that's it. 01:02:55.290 --> 01:02:57.707 align:middle line:90% You have a finding right there. 01:02:57.707 --> 01:02:58.790 align:middle line:90% Because the data is there. 01:02:58.790 --> 01:03:02.883 align:middle line:84% Hardly anyone knows how to intersect these things at all. 01:03:02.883 --> 01:03:04.550 align:middle line:84% And so what I'm going to talk about here 01:03:04.550 --> 01:03:07.527 align:middle line:84% for the last few minutes is this concept of discovery portal. 01:03:07.527 --> 01:03:08.610 align:middle line:90% So what do I mean by that? 01:03:08.610 --> 01:03:09.650 align:middle line:90% Let me skip this part. 01:03:09.650 --> 01:03:15.830 align:middle line:90% 01:03:15.830 --> 01:03:18.560 align:middle line:84% We start with in-vivo, experiments 01:03:18.560 --> 01:03:20.150 align:middle line:90% done in humans and mice. 01:03:20.150 --> 01:03:21.860 align:middle line:84% Then we had in vitro and cell culture. 01:03:21.860 --> 01:03:23.930 align:middle line:90% Now we have NHTTP. 01:03:23.930 --> 01:03:25.280 align:middle line:90% So here's an idea. 01:03:25.280 --> 01:03:28.460 align:middle line:84% Here's how you can go from an actual thought 01:03:28.460 --> 01:03:30.800 align:middle line:84% to a pretty well established hypothesis 01:03:30.800 --> 01:03:32.510 align:middle line:84% without even leaving your web browser, 01:03:32.510 --> 01:03:35.100 align:middle line:84% because all of these tools are available today. 01:03:35.100 --> 01:03:36.740 align:middle line:84% Now for the next maybe 15 slides, 01:03:36.740 --> 01:03:38.370 align:middle line:84% I'm going to flip through them pretty quickly. 01:03:38.370 --> 01:03:40.453 align:middle line:84% I'm going to cover a number of different databases 01:03:40.453 --> 01:03:43.530 align:middle line:84% where you could get all sorts of inspirational ideas from. 01:03:43.530 --> 01:03:46.550 align:middle line:84% So I'm going to start with an interest, move to a rat model. 01:03:46.550 --> 01:03:48.590 align:middle line:84% We're going to cover QTLs while we're there 01:03:48.590 --> 01:03:49.850 align:middle line:90% and physiological findings. 01:03:49.850 --> 01:03:51.380 align:middle line:90% Go to a rat genome map. 01:03:51.380 --> 01:03:53.540 align:middle line:84% Go to multiple species expression data, 01:03:53.540 --> 01:03:55.980 align:middle line:84% and end with the hypothesis here. 01:03:55.980 --> 01:03:56.630 align:middle line:90% This is live. 01:03:56.630 --> 01:03:58.203 align:middle line:84% You could do this yourself today. 01:03:58.203 --> 01:03:59.870 align:middle line:84% We're going to start at a website called 01:03:59.870 --> 01:04:03.110 align:middle line:84% PhysGen. Some of us in the room know about this. 01:04:03.110 --> 01:04:08.870 align:middle line:84% This is at Medical College of Wisconsin. 01:04:08.870 --> 01:04:12.440 align:middle line:84% What these guys do is they have a huge pipeline of things 01:04:12.440 --> 01:04:15.380 align:middle line:84% they do to mice and rats to measure things, 01:04:15.380 --> 01:04:17.150 align:middle line:84% measure all sorts of different quantities. 01:04:17.150 --> 01:04:19.880 align:middle line:84% How much does this rat pee versus that rat? 01:04:19.880 --> 01:04:22.610 align:middle line:84% How much does this mouse breathe compared to this mouse? 01:04:22.610 --> 01:04:24.027 align:middle line:84% Et cetera et cetera, and they have 01:04:24.027 --> 01:04:26.670 align:middle line:84% tables of all of this data publicly available. 01:04:26.670 --> 01:04:29.900 align:middle line:84% So if you go to PhysGen, if you click on Data, 01:04:29.900 --> 01:04:31.310 align:middle line:84% and if you click on Animal Model, 01:04:31.310 --> 01:04:35.210 align:middle line:84% some genotype information for all strains. 01:04:35.210 --> 01:04:40.790 align:middle line:84% Now as I scroll down a little I see a whole bunch of the rats. 01:04:40.790 --> 01:04:43.250 align:middle line:84% Now I might remember, because I went to the Diabetes 01:04:43.250 --> 01:04:45.200 align:middle line:84% Association conference last year, 01:04:45.200 --> 01:04:47.510 align:middle line:84% I remember hearing that's the Brown Norway 01:04:47.510 --> 01:04:49.670 align:middle line:90% has some problem with diabetes. 01:04:49.670 --> 01:04:52.610 align:middle line:84% I might not have even gone to the seminars on the Brown, 01:04:52.610 --> 01:04:55.100 align:middle line:84% but everyone kept talking about Brown Norway, Brown Norway. 01:04:55.100 --> 01:04:57.183 align:middle line:84% I heard a Brown Norway has a problem with disease. 01:04:57.183 --> 01:05:00.285 align:middle line:84% Let's go click on the Brown Norway strain report. 01:05:00.285 --> 01:05:01.160 align:middle line:90% Scroll down a little. 01:05:01.160 --> 01:05:03.290 align:middle line:90% This is the BN, or Brown Norway. 01:05:03.290 --> 01:05:05.180 align:middle line:84% And I can see, yes it does, it has 01:05:05.180 --> 01:05:07.920 align:middle line:90% diabetes on here, diabetes QTL. 01:05:07.920 --> 01:05:10.320 align:middle line:90% So what is a QTL? 01:05:10.320 --> 01:05:13.430 align:middle line:84% So we've got the rat, and it's got its own genome, 01:05:13.430 --> 01:05:15.320 align:middle line:90% like any other creature. 01:05:15.320 --> 01:05:18.180 align:middle line:84% And what a QTL is a quantitative trait loci. 01:05:18.180 --> 01:05:22.730 align:middle line:84% So a particular portion of a particular chromosome 01:05:22.730 --> 01:05:25.460 align:middle line:84% appears to be statistically significantly 01:05:25.460 --> 01:05:28.910 align:middle line:84% associated with a particular phenotypic measurement. 01:05:28.910 --> 01:05:30.810 align:middle line:84% That's a quantitative trait loci. 01:05:30.810 --> 01:05:33.050 align:middle line:84% So in other words, this particular region 01:05:33.050 --> 01:05:36.050 align:middle line:84% which we're going to look at in a second on chromosome 2 01:05:36.050 --> 01:05:40.820 align:middle line:84% is statistically associated with the insulin level, the fasting 01:05:40.820 --> 01:05:43.760 align:middle line:90% insulin level of a rat. 01:05:43.760 --> 01:05:47.168 align:middle line:84% So notice if you have one version of this, 01:05:47.168 --> 01:05:48.960 align:middle line:84% you might have a low fasting insulin level. 01:05:48.960 --> 01:05:51.418 align:middle line:84% If you have another version, you have a high fasting level. 01:05:51.418 --> 01:05:55.692 align:middle line:84% But something in this region of a chromosome is associated. 01:05:55.692 --> 01:05:57.650 align:middle line:84% We're not at the gene level yet, but we're just 01:05:57.650 --> 01:06:00.213 align:middle line:84% in a region of a chromosome as a starting point 01:06:00.213 --> 01:06:01.880 align:middle line:84% and and ending point of this chromosome, 01:06:01.880 --> 01:06:04.790 align:middle line:84% and some piece in the middle is associated. 01:06:04.790 --> 01:06:07.040 align:middle line:84% There's all sorts of other measurements here. 01:06:07.040 --> 01:06:08.737 align:middle line:84% You've got X-ray hypersensitivity. 01:06:08.737 --> 01:06:09.570 align:middle line:90% What does that mean? 01:06:09.570 --> 01:06:13.070 align:middle line:84% Well something on this region of chromosome 1 01:06:13.070 --> 01:06:16.400 align:middle line:84% means you get more sensitive to X-rays. 01:06:16.400 --> 01:06:19.150 align:middle line:84% You can think about how esoteric these measurements are. 01:06:19.150 --> 01:06:20.900 align:middle line:84% If you're interested in radiation effects, 01:06:20.900 --> 01:06:22.442 align:middle line:84% there's your measurement right there. 01:06:22.442 --> 01:06:25.740 align:middle line:84% Arthritis severity, blood pressure, body weight, 01:06:25.740 --> 01:06:27.590 align:middle line:90% blood pressure, blood pressure. 01:06:27.590 --> 01:06:31.520 align:middle line:84% They're all sitting there, waiting to be tapped into. 01:06:31.520 --> 01:06:34.610 align:middle line:84% No one has enough time to look at every single one of these, 01:06:34.610 --> 01:06:36.950 align:middle line:84% but we're mandated by NIH to just put them out 01:06:36.950 --> 01:06:39.465 align:middle line:84% on the internet for people like you to just study. 01:06:39.465 --> 01:06:41.090 align:middle line:84% If you're interested in blood pressure, 01:06:41.090 --> 01:06:44.810 align:middle line:84% there's three QTLs right there that people by chance, people 01:06:44.810 --> 01:06:46.490 align:middle line:84% have probably never even looked at. 01:06:46.490 --> 01:06:50.220 align:middle line:84% Let's click on this particular QTL to see what it looks like. 01:06:50.220 --> 01:06:53.840 align:middle line:84% Here's the name of the QTL, Nidd/gk2. 01:06:53.840 --> 01:06:55.610 align:middle line:90% It's in the Brown Norway. 01:06:55.610 --> 01:06:58.310 align:middle line:84% And here's the marker for this particular peak. 01:06:58.310 --> 01:07:02.060 align:middle line:84% So a marker means this is a unique sequence 01:07:02.060 --> 01:07:04.250 align:middle line:84% in the chromosome that is going to tell me 01:07:04.250 --> 01:07:07.860 align:middle line:84% whether I have the one version of this or the other. 01:07:07.860 --> 01:07:12.890 align:middle line:84% Let me click on that marker, D2Wox23. 01:07:12.890 --> 01:07:15.380 align:middle line:84% And here is the PCR test I would need 01:07:15.380 --> 01:07:18.050 align:middle line:84% to do in the rack to tell me, do I 01:07:18.050 --> 01:07:20.210 align:middle line:84% have the one letter or the other letter? 01:07:20.210 --> 01:07:23.540 align:middle line:84% Or in other words, do I have the high insulin 01:07:23.540 --> 01:07:25.400 align:middle line:84% version or the low insulin version 01:07:25.400 --> 01:07:26.870 align:middle line:90% of this particular peak? 01:07:26.870 --> 01:07:28.130 align:middle line:90% So it even gives you the test. 01:07:28.130 --> 01:07:30.470 align:middle line:84% You could just go order those oligonucleotides, 01:07:30.470 --> 01:07:33.280 align:middle line:84% find a rat that's walking down Longwood Avenue at night, 01:07:33.280 --> 01:07:35.780 align:middle line:84% and just test to see, is this a high insulin or low insulin? 01:07:35.780 --> 01:07:37.760 align:middle line:90% You can do it today. 01:07:37.760 --> 01:07:40.880 align:middle line:84% Now out of dumb luck it just so happens that they've already 01:07:40.880 --> 01:07:42.500 align:middle line:84% done the work, and they've told us 01:07:42.500 --> 01:07:48.000 align:middle line:84% in this region of the chromosome we know there's a gene there. 01:07:48.000 --> 01:07:52.340 align:middle line:84% It doesn't have to be, but there is a gene there, S100A4. 01:07:52.340 --> 01:07:58.013 align:middle line:84% That's a gene that exists in this peak of significance. 01:07:58.013 --> 01:08:00.680 align:middle line:84% Now what we're going to do is go take that piece of information, 01:08:00.680 --> 01:08:02.263 align:middle line:84% that name of that gene, and let's look 01:08:02.263 --> 01:08:04.910 align:middle line:84% more at that QTL region, that region of the chromosome. 01:08:04.910 --> 01:08:07.160 align:middle line:84% We're going to go to National Center for Biotechnology 01:08:07.160 --> 01:08:09.950 align:middle line:84% Information, click on Genomic Biology. 01:08:09.950 --> 01:08:13.010 align:middle line:90% I'm going to click on the rat. 01:08:13.010 --> 01:08:15.800 align:middle line:84% And I'm going to just type in S100A4, the name of that gene 01:08:15.800 --> 01:08:17.899 align:middle line:90% I told us was in that region. 01:08:17.899 --> 01:08:19.729 align:middle line:84% Where is it in the rat chromosome? 01:08:19.729 --> 01:08:22.160 align:middle line:90% Here it is on chromosome 2. 01:08:22.160 --> 01:08:23.090 align:middle line:90% It's S100A4. 01:08:23.090 --> 01:08:29.569 align:middle line:84% If I click on that, if I zoom out a little, here's S100A4, 01:08:29.569 --> 01:08:33.218 align:middle line:84% and here are all the neighboring genes. 01:08:33.218 --> 01:08:34.010 align:middle line:90% That's interesting. 01:08:34.010 --> 01:08:35.802 align:middle line:84% There's a bunch of other interesting genes. 01:08:35.802 --> 01:08:39.229 align:middle line:84% First, S100A4, there's A3, A6, A8, A9. 01:08:39.229 --> 01:08:43.345 align:middle line:84% This gene is one in many of a gene family. 01:08:43.345 --> 01:08:44.720 align:middle line:84% So how new genes are often formed 01:08:44.720 --> 01:08:46.850 align:middle line:84% is through duplication events like this. 01:08:46.850 --> 01:08:48.380 align:middle line:84% But to me more interestingly I see 01:08:48.380 --> 01:08:50.479 align:middle line:84% things like natriuretic peptide receptor. 01:08:50.479 --> 01:08:53.517 align:middle line:84% Natriuretic peptide is made in a number of different sources, 01:08:53.517 --> 01:08:54.350 align:middle line:90% including the heart. 01:08:54.350 --> 01:08:57.319 align:middle line:84% It has to do with blood pressure and blood volume regulation. 01:08:57.319 --> 01:08:58.652 align:middle line:90% That's an interesting hormone. 01:08:58.652 --> 01:08:59.569 align:middle line:90% Let's see some others. 01:08:59.569 --> 01:09:02.300 align:middle line:84% Interleukin 6 there, might be some interesting things there. 01:09:02.300 --> 01:09:06.470 align:middle line:84% But let's just keep running with this S100A4 here. 01:09:06.470 --> 01:09:08.300 align:middle line:84% Now what I'm going to do is, I'm going 01:09:08.300 --> 01:09:12.080 align:middle line:84% to look at where is this gene expressed. 01:09:12.080 --> 01:09:14.510 align:middle line:90% I got it in a statistical peak. 01:09:14.510 --> 01:09:17.069 align:middle line:84% I've seen what its neighbors are in the genome. 01:09:17.069 --> 01:09:20.270 align:middle line:84% Now I want to know, why is it that this particular gene 01:09:20.270 --> 01:09:23.215 align:middle line:84% is associated with diabetes or insulin level? 01:09:23.215 --> 01:09:23.840 align:middle line:90% Well let's see. 01:09:23.840 --> 01:09:25.880 align:middle line:84% Where is that gene expressed in a whole bunch 01:09:25.880 --> 01:09:27.050 align:middle line:90% of different samples? 01:09:27.050 --> 01:09:28.700 align:middle line:84% This is one particular search engine 01:09:28.700 --> 01:09:30.658 align:middle line:84% where you can just type in the name of the gene 01:09:30.658 --> 01:09:32.630 align:middle line:84% and see where is this gene expressed in about 01:09:32.630 --> 01:09:34.520 align:middle line:90% 1,200 different samples. 01:09:34.520 --> 01:09:38.029 align:middle line:84% This is specifically cardiovascular. 01:09:38.029 --> 01:09:40.340 align:middle line:84% I type in S100A4, it says, which do you mean? 01:09:40.340 --> 01:09:41.927 align:middle line:84% The human, the mouse, or the rat? 01:09:41.927 --> 01:09:43.760 align:middle line:84% Well we started with the rat, but let's just 01:09:43.760 --> 01:09:45.560 align:middle line:84% see where the human ones are expressed. 01:09:45.560 --> 01:09:47.750 align:middle line:90% Click on that. 01:09:47.750 --> 01:09:50.960 align:middle line:84% It's measured in 10 microarrays by Johns Hopkins. 01:09:50.960 --> 01:09:51.859 align:middle line:90% That's not that many. 01:09:51.859 --> 01:09:53.010 align:middle line:90% Let's go back. 01:09:53.010 --> 01:09:54.630 align:middle line:90% Let's look at the mouse one. 01:09:54.630 --> 01:09:57.540 align:middle line:84% Mouse is measured by 142 by this group, 100 by this group, 01:09:57.540 --> 01:09:58.770 align:middle line:90% but 84 by this group. 01:09:58.770 --> 01:10:00.990 align:middle line:90% Let's look at the 84 first. 01:10:00.990 --> 01:10:03.320 align:middle line:90% And we can see that top samples. 01:10:03.320 --> 01:10:05.450 align:middle line:90% So this is the expression level. 01:10:05.450 --> 01:10:08.090 align:middle line:84% This is the percentile of that gene on that chip, 01:10:08.090 --> 01:10:09.200 align:middle line:90% on each chip. 01:10:09.200 --> 01:10:12.620 align:middle line:84% Muscle, muscle, muscle, muscle, muscle, muscle, muscle, muscle. 01:10:12.620 --> 01:10:14.780 align:middle line:84% And at least for Johns Hopkins, this gene 01:10:14.780 --> 01:10:17.510 align:middle line:84% is most highly expressed in their muscle samples, 01:10:17.510 --> 01:10:19.337 align:middle line:84% specifically regenerating muscle. 01:10:19.337 --> 01:10:21.170 align:middle line:84% We can look to see exactly what sample there 01:10:21.170 --> 01:10:22.640 align:middle line:90% is, but let's go on. 01:10:22.640 --> 01:10:25.220 align:middle line:84% Let's hit the Back button, and let's look at cardiogenomics, 01:10:25.220 --> 01:10:27.770 align:middle line:84% so that's these guys at the Beth Israel. 01:10:27.770 --> 01:10:31.590 align:middle line:84% And again, it's highly expressed in banded smooth muscle. 01:10:31.590 --> 01:10:33.410 align:middle line:90% Now it says band here. 01:10:33.410 --> 01:10:35.900 align:middle line:84% That says band 48 hours, band 48 hours, 24 hours. 01:10:35.900 --> 01:10:37.100 align:middle line:90% What exactly did they do? 01:10:37.100 --> 01:10:39.170 align:middle line:90% What exactly are these samples? 01:10:39.170 --> 01:10:41.843 align:middle line:84% You click on Overview, it takes you to their website. 01:10:41.843 --> 01:10:43.760 align:middle line:84% It tells you what was the experiment they did. 01:10:43.760 --> 01:10:46.730 align:middle line:84% They actually did an experiment where they took these mice, 01:10:46.730 --> 01:10:49.160 align:middle line:84% they opened them up, and they pull a little rubber band 01:10:49.160 --> 01:10:51.643 align:middle line:90% on their aortas. 01:10:51.643 --> 01:10:52.310 align:middle line:90% In half of them. 01:10:52.310 --> 01:10:53.840 align:middle line:84% The other half, they just sewed them up again. 01:10:53.840 --> 01:10:54.848 align:middle line:90% They didn't do anything. 01:10:54.848 --> 01:10:56.390 align:middle line:84% And so they made a list of genes that 01:10:56.390 --> 01:11:01.460 align:middle line:84% go up in the heart as a response to this pressure overload. 01:11:01.460 --> 01:11:02.990 align:middle line:90% The genes are sitting there. 01:11:02.990 --> 01:11:05.510 align:middle line:84% I can guarantee you again, not a single paper 01:11:05.510 --> 01:11:07.430 align:middle line:90% has been published on this list. 01:11:07.430 --> 01:11:09.650 align:middle line:84% Because we're so busy creating more data, 01:11:09.650 --> 01:11:13.250 align:middle line:84% we have not had time to publish anything on this data. 01:11:13.250 --> 01:11:15.830 align:middle line:84% So they were trying to study pressure overload induced 01:11:15.830 --> 01:11:16.940 align:middle line:90% cardiac hypertrophy. 01:11:16.940 --> 01:11:18.590 align:middle line:90% So now here's the hypothesis. 01:11:18.590 --> 01:11:21.020 align:middle line:84% I started with the Brown Norway rat, 01:11:21.020 --> 01:11:22.640 align:middle line:84% and the Brown Norway rat has signs 01:11:22.640 --> 01:11:25.940 align:middle line:84% of diabetes because of this gene, S100A4, 01:11:25.940 --> 01:11:29.780 align:middle line:84% because S100A4 was under that significant peak. 01:11:29.780 --> 01:11:32.330 align:middle line:84% But I think it's because S100A4 is expressed 01:11:32.330 --> 01:11:34.880 align:middle line:84% in muscle under a variety of different conditions, 01:11:34.880 --> 01:11:38.338 align:middle line:84% and I know muscle is an insulin target tissue. 01:11:38.338 --> 01:11:40.880 align:middle line:84% So now I can get some clues to what experiment I can do next. 01:11:40.880 --> 01:11:43.892 align:middle line:84% Maybe I should go look up muscle cells for S100A4. 01:11:43.892 --> 01:11:45.350 align:middle line:84% If you actually do PubMed searches, 01:11:45.350 --> 01:11:47.540 align:middle line:84% you can see S100A4 is associated-- 01:11:47.540 --> 01:11:50.405 align:middle line:84% at least it's being studied in diabetes today. 01:11:50.405 --> 01:11:53.900 align:middle line:84% Now I just picked the first peak I saw on that website. 01:11:53.900 --> 01:11:56.240 align:middle line:84% You can go back and find any number of new genes 01:11:56.240 --> 01:11:57.710 align:middle line:84% there and come up with hypotheses, 01:11:57.710 --> 01:12:00.763 align:middle line:84% and we didn't even leave the web browser. 01:12:00.763 --> 01:12:03.180 align:middle line:84% You didn't even have to download a file to do any of this. 01:12:03.180 --> 01:12:04.523 align:middle line:90% So you could do this today. 01:12:04.523 --> 01:12:06.440 align:middle line:84% You start just surfing from one of these sites 01:12:06.440 --> 01:12:08.940 align:middle line:84% to another from something you might have heard down the hall 01:12:08.940 --> 01:12:10.250 align:middle line:90% to an actual hypothesis. 01:12:10.250 --> 01:12:16.780 align:middle line:90% 01:12:16.780 --> 01:12:17.960 align:middle line:90% We have about 10 minutes. 01:12:17.960 --> 01:12:19.630 align:middle line:84% I'm just going to call it quits now 01:12:19.630 --> 01:12:24.785 align:middle line:84% to see if you have any questions or thoughts. 01:12:24.785 --> 01:12:30.830 align:middle line:90% 01:12:30.830 --> 01:12:34.835 align:middle line:84% AUDIENCE: So at what point with these various websites 01:12:34.835 --> 01:12:36.710 align:middle line:84% that have the data, what place does that data 01:12:36.710 --> 01:12:37.680 align:middle line:90% get put on the website? 01:12:37.680 --> 01:12:40.700 align:middle line:84% Are they after publication versus [INAUDIBLE] sort 01:12:40.700 --> 01:12:41.200 align:middle line:90% of like-- 01:12:41.200 --> 01:12:42.283 align:middle line:90% ATUL BUTTE: Good question. 01:12:42.283 --> 01:12:44.710 align:middle line:84% AUDIENCE: [INAUDIBLE] structures [INTERPOSING VOICES] 01:12:44.710 --> 01:12:45.835 align:middle line:90% ATUL BUTTE: Great question. 01:12:45.835 --> 01:12:48.430 align:middle line:84% So NCBI GEO, you could submit data to them, 01:12:48.430 --> 01:12:50.500 align:middle line:84% and then they'll keep it hidden until you 01:12:50.500 --> 01:12:53.140 align:middle line:84% say it's been published, and then they could turn it on. 01:12:53.140 --> 01:12:55.610 align:middle line:84% These other personal websites like the Whitehead one, 01:12:55.610 --> 01:12:58.300 align:middle line:84% Stanford, they typically appear after publication. 01:12:58.300 --> 01:13:02.410 align:middle line:84% For these other websites in this PGA, the cardiovascular data 01:13:02.410 --> 01:13:04.240 align:middle line:84% I keep showing you, that data appears 01:13:04.240 --> 01:13:07.990 align:middle line:84% within 60 days of its creation, regardless of publication. 01:13:07.990 --> 01:13:12.700 align:middle line:84% NIH is more and more mandating that to be the model now. 01:13:12.700 --> 01:13:15.580 align:middle line:84% Because if they're paying for all of this genomic work, 01:13:15.580 --> 01:13:18.940 align:middle line:84% they don't want just one lab to really be able to run with it, 01:13:18.940 --> 01:13:21.670 align:middle line:84% because there's too much data for any one lab to deal with. 01:13:21.670 --> 01:13:25.540 align:middle line:84% So for some of their grants they actually insist 01:13:25.540 --> 01:13:30.040 align:middle line:84% that this data be made available well before publication. 01:13:30.040 --> 01:13:32.350 align:middle line:84% So right now on cardiovascular and lung 01:13:32.350 --> 01:13:36.320 align:middle line:84% disease, and a little bit of sleep as well, 01:13:36.320 --> 01:13:39.670 align:middle line:84% you can get a lot of data, 1,500 arrays. 01:13:39.670 --> 01:13:43.000 align:middle line:90% AUDIENCE: [INAUDIBLE] 01:13:43.000 --> 01:13:45.488 align:middle line:84% ATUL BUTTE: Yep, these PGA- these Programs and Genomic 01:13:45.488 --> 01:13:46.030 align:middle line:90% Applications. 01:13:46.030 --> 01:13:48.370 align:middle line:84% PGA was funded by the Heart, Lung, and Blood Institute. 01:13:48.370 --> 01:13:51.290 align:middle line:84% That in particular has a lot of data that's already available 01:13:51.290 --> 01:13:53.730 align:middle line:90% that hasn't been published on. 01:13:53.730 --> 01:13:57.148 align:middle line:90%