WEBVTT 00:00:00.000 --> 00:00:03.240 align:middle line:84% ALBERTO RIVA: Alberto Riva, I'm an instructor at CHB. 00:00:03.240 --> 00:00:08.430 align:middle line:84% I'm going to talk to you today about the most important 00:00:08.430 --> 00:00:14.902 align:middle line:84% resources for finding and using biomedical information, 00:00:14.902 --> 00:00:16.860 align:middle line:84% especially information connected with the study 00:00:16.860 --> 00:00:20.110 align:middle line:90% of the human genome. 00:00:20.110 --> 00:00:22.980 align:middle line:84% So this is going to be something probably slightly 00:00:22.980 --> 00:00:26.550 align:middle line:84% different from what you've heard so far, which concentrate more 00:00:26.550 --> 00:00:31.320 align:middle line:84% on actual locations, and most of them 00:00:31.320 --> 00:00:33.500 align:middle line:84% will be websites, where you can find information, 00:00:33.500 --> 00:00:36.540 align:middle line:84% and to talk about how this information is stored 00:00:36.540 --> 00:00:38.850 align:middle line:84% and represented, how it's accessible, 00:00:38.850 --> 00:00:42.900 align:middle line:90% and what it can be used for. 00:00:42.900 --> 00:00:46.440 align:middle line:84% So you're going to see a long list of references 00:00:46.440 --> 00:00:48.255 align:middle line:90% to sites, websites, with URLs. 00:00:48.255 --> 00:00:50.130 align:middle line:84% Don't worry if you can't remember all of them 00:00:50.130 --> 00:00:52.380 align:middle line:84% because, of course, I'm going to distribute the slides 00:00:52.380 --> 00:00:55.510 align:middle line:84% and will be easier to just look them up. 00:00:55.510 --> 00:01:03.800 align:middle line:84% So I'm going to start with something that you probably 00:01:03.800 --> 00:01:06.790 align:middle line:90% heard about many times before. 00:01:06.790 --> 00:01:12.305 align:middle line:84% The so-called central dogma of molecular biology, as you know, 00:01:12.305 --> 00:01:16.130 align:middle line:84% almost all our cells contain DNA in their nucleus-- 00:01:16.130 --> 00:01:19.900 align:middle line:84% DNA is the molecule that encodes information-- 00:01:19.900 --> 00:01:23.120 align:middle line:84% for at least from the purposes of this presentation, 00:01:23.120 --> 00:01:24.970 align:middle line:84% this is what we're interested in. 00:01:24.970 --> 00:01:29.350 align:middle line:84% And this information is transcribed 00:01:29.350 --> 00:01:32.230 align:middle line:84% into our name molecules, that then exit the nucleus 00:01:32.230 --> 00:01:35.380 align:middle line:90% under the form of mRNA. 00:01:35.380 --> 00:01:37.900 align:middle line:84% MRNA is then translated into proteins. 00:01:37.900 --> 00:01:44.560 align:middle line:84% And proteins are ultimately what is responsible for essentially 00:01:44.560 --> 00:01:48.550 align:middle line:84% all the external manifestations, all the observable properties 00:01:48.550 --> 00:01:51.910 align:middle line:90% of our biology. 00:01:51.910 --> 00:01:55.480 align:middle line:84% So I'm talking about things like metabolism, general physiology, 00:01:55.480 --> 00:01:59.860 align:middle line:84% diversity between individuals, diseases, drug response. 00:01:59.860 --> 00:02:02.650 align:middle line:90% 00:02:02.650 --> 00:02:05.110 align:middle line:84% These are all, n some way or another, 00:02:05.110 --> 00:02:10.690 align:middle line:84% due to the different proteins that act within our cells 00:02:10.690 --> 00:02:12.170 align:middle line:90% and outside our cells. 00:02:12.170 --> 00:02:17.485 align:middle line:84% So we have names, of course, for the extremes of the spectrum. 00:02:17.485 --> 00:02:20.380 align:middle line:90% 00:02:20.380 --> 00:02:25.690 align:middle line:84% We call genotype-- this information 00:02:25.690 --> 00:02:27.530 align:middle line:90% is encoded in the DNA. 00:02:27.530 --> 00:02:29.320 align:middle line:84% And on the other hand, we call phenotype 00:02:29.320 --> 00:02:31.960 align:middle line:84% whatever is at the other extreme of the spectrum, anything 00:02:31.960 --> 00:02:35.395 align:middle line:84% we can observe, we can measure, from the outside. 00:02:35.395 --> 00:02:39.090 align:middle line:90% 00:02:39.090 --> 00:02:45.100 align:middle line:84% So what I'm going to try to show you is that, as you move 00:02:45.100 --> 00:02:46.850 align:middle line:84% from one end of the spectrum to the other, 00:02:46.850 --> 00:02:49.760 align:middle line:84% you're going to encounter very different forms of information, 00:02:49.760 --> 00:02:50.370 align:middle line:90% of data. 00:02:50.370 --> 00:02:53.810 align:middle line:84% And each one has its own specific nature 00:02:53.810 --> 00:02:57.050 align:middle line:84% and function, and needs to be treated with different tools. 00:02:57.050 --> 00:02:59.880 align:middle line:84% And needs to be represented in different ways. 00:02:59.880 --> 00:03:05.520 align:middle line:84% So essentially these are the questions 00:03:05.520 --> 00:03:07.040 align:middle line:90% that I'm going to answer. 00:03:07.040 --> 00:03:08.870 align:middle line:84% How is all this information represented? 00:03:08.870 --> 00:03:12.320 align:middle line:84% What are the different ways that we can store and describe 00:03:12.320 --> 00:03:13.400 align:middle line:90% this information? 00:03:13.400 --> 00:03:14.870 align:middle line:84% Where does it come from, where is 00:03:14.870 --> 00:03:17.510 align:middle line:84% it stored, how do we find retrieve and use it? 00:03:17.510 --> 00:03:20.230 align:middle line:90% 00:03:20.230 --> 00:03:24.100 align:middle line:84% So we've talked about the two ends of the spectrum. 00:03:24.100 --> 00:03:29.970 align:middle line:84% So there are some very deep differences 00:03:29.970 --> 00:03:31.680 align:middle line:84% between the kind of information you 00:03:31.680 --> 00:03:35.483 align:middle line:84% find when you're talking about the genotype, 00:03:35.483 --> 00:03:37.150 align:middle line:84% and when we're talking about phenotypes. 00:03:37.150 --> 00:03:38.700 align:middle line:84% For example, the genotype is digital, 00:03:38.700 --> 00:03:40.440 align:middle line:84% because each base pair in our DNA 00:03:40.440 --> 00:03:44.298 align:middle line:84% can be exactly represented using one of four symbols, A, P, G, 00:03:44.298 --> 00:03:46.590 align:middle line:84% C. Then you can also think about insertions, deletions, 00:03:46.590 --> 00:03:47.090 align:middle line:90% and so on. 00:03:47.090 --> 00:03:50.910 align:middle line:84% But essentially, using a small number of symbols, 00:03:50.910 --> 00:03:54.780 align:middle line:84% you can provide an exact representation of our genome, 00:03:54.780 --> 00:03:59.560 align:middle line:84% of all the 3 billion base pairs that compose our DNA. 00:03:59.560 --> 00:04:04.500 align:middle line:84% On the other hand, the phenotype is, say, analog. 00:04:04.500 --> 00:04:08.208 align:middle line:84% Because most phenotypes are qualitative in nature. 00:04:08.208 --> 00:04:10.000 align:middle line:84% They cannot be measured exactly or precise, 00:04:10.000 --> 00:04:14.519 align:middle line:84% they cannot even be defined precisely in most cases. 00:04:14.519 --> 00:04:16.019 align:middle line:84% You always have to take into account 00:04:16.019 --> 00:04:17.490 align:middle line:84% the effect of environmental factors 00:04:17.490 --> 00:04:23.400 align:middle line:84% that, again, are very hard to describe in a quantitative way. 00:04:23.400 --> 00:04:28.050 align:middle line:84% At the root of all this is that one of the biggest problems 00:04:28.050 --> 00:04:30.090 align:middle line:84% in the study of protein is the fact 00:04:30.090 --> 00:04:32.700 align:middle line:84% that the proteins are not uniquely determined 00:04:32.700 --> 00:04:34.957 align:middle line:90% by their sequence. 00:04:34.957 --> 00:04:36.540 align:middle line:84% For DNA, you just look at the sequence 00:04:36.540 --> 00:04:38.010 align:middle line:84% and you know essentially all that there 00:04:38.010 --> 00:04:38.885 align:middle line:90% is to know about DNA. 00:04:38.885 --> 00:04:41.877 align:middle line:84% For proteins, you cannot look at the subsequence of a protein 00:04:41.877 --> 00:04:44.460 align:middle line:84% and understand just by looking at it what the protein is going 00:04:44.460 --> 00:04:44.960 align:middle line:90% to do. 00:04:44.960 --> 00:04:48.243 align:middle line:90% Not even how it's going to be-- 00:04:48.243 --> 00:04:50.160 align:middle line:84% not even what it's three-dimensional structure 00:04:50.160 --> 00:04:50.785 align:middle line:90% is going to be. 00:04:50.785 --> 00:04:52.082 align:middle line:90% That's difficult enough. 00:04:52.082 --> 00:04:53.790 align:middle line:84% Then understanding what the protein does, 00:04:53.790 --> 00:04:55.830 align:middle line:84% just by looking at the sequence, is still 00:04:55.830 --> 00:04:57.750 align:middle line:90% very far from being feasible. 00:04:57.750 --> 00:05:00.950 align:middle line:90% 00:05:00.950 --> 00:05:02.700 align:middle line:84% On the other hand, it's interesting to see 00:05:02.700 --> 00:05:05.070 align:middle line:84% that our knowledge of these things 00:05:05.070 --> 00:05:07.140 align:middle line:84% has progressed in the opposite direction. 00:05:07.140 --> 00:05:09.690 align:middle line:84% Because obviously, it's much easier to observe a phenotype 00:05:09.690 --> 00:05:11.350 align:middle line:84% than to observe DNA in ourselves. 00:05:11.350 --> 00:05:14.250 align:middle line:84% So the first thought is of inherited traits date 00:05:14.250 --> 00:05:17.490 align:middle line:90% back to Mendel in 1866. 00:05:17.490 --> 00:05:24.820 align:middle line:84% And DNA was discovered, more or less, the same years. 00:05:24.820 --> 00:05:26.778 align:middle line:84% But at the time, nobody had any idea 00:05:26.778 --> 00:05:29.070 align:middle line:84% that there was any connection between these two things, 00:05:29.070 --> 00:05:32.910 align:middle line:84% between the DNA and inherited traits. 00:05:32.910 --> 00:05:42.810 align:middle line:84% It took over 80 years for this concept to be proven. 00:05:42.810 --> 00:05:48.270 align:middle line:84% So the definite proof that genes are made of DNA 00:05:48.270 --> 00:05:50.410 align:middle line:90% dates back to 1952. 00:05:50.410 --> 00:05:56.430 align:middle line:84% After that, progress was faster because the elucidation 00:05:56.430 --> 00:06:00.000 align:middle line:84% of the structure of DNA, and the DNA replication mechanism, 00:06:00.000 --> 00:06:02.700 align:middle line:90% came one year later. 00:06:02.700 --> 00:06:04.860 align:middle line:84% Then the geniculate code was deciphered 00:06:04.860 --> 00:06:08.215 align:middle line:90% between 1961 and 1966. 00:06:08.215 --> 00:06:09.840 align:middle line:84% Something that we now take for granted, 00:06:09.840 --> 00:06:14.790 align:middle line:84% like the discovery of introns, only happened in 1977. 00:06:14.790 --> 00:06:17.460 align:middle line:84% And finally, the Human Genome Project, 00:06:17.460 --> 00:06:26.060 align:middle line:84% that was officially declared a success last year, 00:06:26.060 --> 00:06:29.300 align:middle line:84% brought us to the point where we now know the exact base pair 00:06:29.300 --> 00:06:32.010 align:middle line:90% sequence of our genome. 00:06:32.010 --> 00:06:36.028 align:middle line:84% So we now know with a sufficient degree of certainty-- 00:06:36.028 --> 00:06:37.820 align:middle line:84% we're going to talk about this more later-- 00:06:37.820 --> 00:06:40.520 align:middle line:84% we know the exact base pair composition 00:06:40.520 --> 00:06:42.170 align:middle line:90% of the human genome. 00:06:42.170 --> 00:06:44.550 align:middle line:84% And also several other genomes, but of course, 00:06:44.550 --> 00:06:46.717 align:middle line:84% the human one is the one we're most interested in. 00:06:46.717 --> 00:06:50.200 align:middle line:90% 00:06:50.200 --> 00:06:52.180 align:middle line:84% Going back again from genotype to phenotype, 00:06:52.180 --> 00:06:53.740 align:middle line:90% there is another thing to note. 00:06:53.740 --> 00:06:56.560 align:middle line:84% I've just said that we now have the complete sequence 00:06:56.560 --> 00:06:58.990 align:middle line:84% for our genome, but of course, this is an approximation. 00:06:58.990 --> 00:07:03.520 align:middle line:84% It's an abstraction, actually, because even if we're all 00:07:03.520 --> 00:07:06.550 align:middle line:84% human beings, there are no two human beings that 00:07:06.550 --> 00:07:07.720 align:middle line:90% are exactly the same. 00:07:07.720 --> 00:07:09.760 align:middle line:84% And this is a consequence of the fact 00:07:09.760 --> 00:07:15.960 align:middle line:84% that there are differences between the DNA of two 00:07:15.960 --> 00:07:17.220 align:middle line:90% any human beings. 00:07:17.220 --> 00:07:20.310 align:middle line:84% These differences are due to polymorphisms, 00:07:20.310 --> 00:07:22.020 align:middle line:84% like single nucleotide polymorphisms, so 00:07:22.020 --> 00:07:24.780 align:middle line:84% locations which, instead of having 00:07:24.780 --> 00:07:28.290 align:middle line:84% the base that everybody else has, you have a different base. 00:07:28.290 --> 00:07:30.780 align:middle line:84% Microsatellites, repeats, insertions, deletions, 00:07:30.780 --> 00:07:32.280 align:middle line:84% translocations, these are all things 00:07:32.280 --> 00:07:34.350 align:middle line:84% that can happen to your DNA sequence that 00:07:34.350 --> 00:07:42.720 align:middle line:84% can modify in ways that, of course, are not enough to turn 00:07:42.720 --> 00:07:45.270 align:middle line:84% you into another animal-- you're still a human being-- 00:07:45.270 --> 00:07:46.980 align:middle line:84% but your DNA sequence is slightly 00:07:46.980 --> 00:07:52.470 align:middle line:84% different from the sequence of any other human being. 00:07:52.470 --> 00:07:55.810 align:middle line:84% So the other is, one of these polymorphisms, every 1,000 00:07:55.810 --> 00:07:56.310 align:middle line:90% bases. 00:07:56.310 --> 00:07:58.420 align:middle line:84% So if you think we have 3 billion base pairs, 00:07:58.420 --> 00:08:03.865 align:middle line:84% it adds up to a very large number of differences. 00:08:03.865 --> 00:08:06.240 align:middle line:84% Which means that when you study the human genome, I mean, 00:08:06.240 --> 00:08:09.490 align:middle line:84% we now have the sequence of the human genome. 00:08:09.490 --> 00:08:13.110 align:middle line:84% But then if you go and look at one individual, 00:08:13.110 --> 00:08:16.493 align:middle line:84% you're not going to find that his DNA matches 00:08:16.493 --> 00:08:17.910 align:middle line:84% exactly the sequence that you find 00:08:17.910 --> 00:08:20.310 align:middle line:90% in the human genome databases. 00:08:20.310 --> 00:08:22.410 align:middle line:84% You're going to find approximately one difference 00:08:22.410 --> 00:08:23.610 align:middle line:90% every 1,000 bases. 00:08:23.610 --> 00:08:26.255 align:middle line:84% And understanding what these differences do and mean, 00:08:26.255 --> 00:08:27.630 align:middle line:84% and what is their consequence, is 00:08:27.630 --> 00:08:29.640 align:middle line:84% one of the most interesting problems 00:08:29.640 --> 00:08:34.049 align:middle line:84% in current bioinformatics and molecular biology, 00:08:34.049 --> 00:08:37.350 align:middle line:84% because now we finally have the tools of looking at our genome 00:08:37.350 --> 00:08:38.460 align:middle line:90% with this level of detail. 00:08:38.460 --> 00:08:40.140 align:middle line:84% We can look at individual base pairs and we can see, 00:08:40.140 --> 00:08:42.570 align:middle line:84% well, there should be an A here, and instead, we 00:08:42.570 --> 00:08:46.270 align:middle line:84% have a C. Does that cause a problem? 00:08:46.270 --> 00:08:50.910 align:middle line:84% So again, we're going to go back to this soon. 00:08:50.910 --> 00:08:52.890 align:middle line:84% And the same thing happens for phenotypes, 00:08:52.890 --> 00:08:54.930 align:middle line:84% although in a slightly different ways. 00:08:54.930 --> 00:08:56.530 align:middle line:84% Phenotypes are generalizations, too. 00:08:56.530 --> 00:08:58.920 align:middle line:84% So when we talk about things like species. 00:08:58.920 --> 00:09:00.557 align:middle line:84% Again, I said, we're all human beings, 00:09:00.557 --> 00:09:02.890 align:middle line:84% but we're all different from a geniculate point of view. 00:09:02.890 --> 00:09:05.790 align:middle line:84% So it means that putting all of us 00:09:05.790 --> 00:09:11.490 align:middle line:84% together into one big group, in one species, 00:09:11.490 --> 00:09:13.050 align:middle line:90% is a generalization, of course. 00:09:13.050 --> 00:09:15.330 align:middle line:84% And even going down to ethnicity, 00:09:15.330 --> 00:09:17.820 align:middle line:84% or even a concept like disease, these are generalizations, 00:09:17.820 --> 00:09:24.030 align:middle line:84% because these are concepts that cannot be defined in a precise, 00:09:24.030 --> 00:09:24.825 align:middle line:90% formal way. 00:09:24.825 --> 00:09:27.360 align:middle line:90% 00:09:27.360 --> 00:09:34.860 align:middle line:84% So we will see as we go forward that we're going to encounter 00:09:34.860 --> 00:09:38.740 align:middle line:90% very different forms of data. 00:09:38.740 --> 00:09:40.500 align:middle line:84% And we're going to need different methods 00:09:40.500 --> 00:09:42.990 align:middle line:84% to manipulate this data, according to what 00:09:42.990 --> 00:09:44.760 align:middle line:90% the purpose of our work is. 00:09:44.760 --> 00:09:47.130 align:middle line:84% So just to make this clear, if we 00:09:47.130 --> 00:09:51.030 align:middle line:84% are working at the level of DNA, then 00:09:51.030 --> 00:09:55.470 align:middle line:84% the typical operations we might be interested in doing 00:09:55.470 --> 00:09:56.970 align:middle line:84% are, for example, sequence matching. 00:09:56.970 --> 00:10:01.680 align:middle line:84% So to understand if certain stretch of a sequence 00:10:01.680 --> 00:10:05.108 align:middle line:84% matches anything else that you've seen before. 00:10:05.108 --> 00:10:07.650 align:middle line:84% So this is useful, for example, when you discover a new gene, 00:10:07.650 --> 00:10:10.350 align:middle line:84% you want to know, first of all, if it's really a new gene, 00:10:10.350 --> 00:10:12.640 align:middle line:84% or if it's already been seen somewhere else. 00:10:12.640 --> 00:10:14.250 align:middle line:84% And if it's a new gene, you would 00:10:14.250 --> 00:10:17.147 align:middle line:84% like to understand-- have an idea of what it does. 00:10:17.147 --> 00:10:19.230 align:middle line:84% And if you find a similarity between your new gene 00:10:19.230 --> 00:10:20.688 align:middle line:84% and something that's already known, 00:10:20.688 --> 00:10:23.665 align:middle line:84% that can give a lot of information. 00:10:23.665 --> 00:10:25.290 align:middle line:84% We're talking about discovering genes-- 00:10:25.290 --> 00:10:29.627 align:middle line:84% finding genes in a DNA sequence is not trivial. 00:10:29.627 --> 00:10:30.960 align:middle line:90% There are programs that do this. 00:10:30.960 --> 00:10:32.585 align:middle line:84% They just look at the sequence and they 00:10:32.585 --> 00:10:34.110 align:middle line:84% find locations in the DNA sequence 00:10:34.110 --> 00:10:36.090 align:middle line:90% that might contain genes. 00:10:36.090 --> 00:10:38.670 align:middle line:84% And I'm not going to go into details on this, 00:10:38.670 --> 00:10:41.220 align:middle line:84% but there are various reasons why 00:10:41.220 --> 00:10:43.920 align:middle line:84% this is a pretty complex thing to do 00:10:43.920 --> 00:10:47.460 align:middle line:84% from a computational point of view. 00:10:47.460 --> 00:10:53.940 align:middle line:84% Homology searches-- again, these refer 00:10:53.940 --> 00:10:57.720 align:middle line:84% to looking for similarities between DNA sequences 00:10:57.720 --> 00:10:58.780 align:middle line:90% in different organisms. 00:10:58.780 --> 00:11:04.290 align:middle line:84% So if you discover the function of a certain gene in the mouse, 00:11:04.290 --> 00:11:06.620 align:middle line:84% for example, you would like to if it 00:11:06.620 --> 00:11:07.870 align:middle line:90% does the same thing in humans. 00:11:07.870 --> 00:11:10.170 align:middle line:84% So again, if you find a high degree of similarity 00:11:10.170 --> 00:11:13.880 align:middle line:84% between the two genes, you can hypothesize 00:11:13.880 --> 00:11:16.005 align:middle line:84% that they are also going to have the same function. 00:11:16.005 --> 00:11:19.445 align:middle line:90% 00:11:19.445 --> 00:11:20.820 align:middle line:84% We've talked about polymorphisms, 00:11:20.820 --> 00:11:25.920 align:middle line:84% so another pretty common cooperation 00:11:25.920 --> 00:11:28.120 align:middle line:84% is performing DNA sequences, is SNP detection. 00:11:28.120 --> 00:11:30.240 align:middle line:84% So if you sequence the same stretch of DNA 00:11:30.240 --> 00:11:32.250 align:middle line:84% from a certain number of different individuals, 00:11:32.250 --> 00:11:33.575 align:middle line:90% then you can compare them. 00:11:33.575 --> 00:11:35.700 align:middle line:84% And you're going to find that most of the locations 00:11:35.700 --> 00:11:38.010 align:middle line:84% are the same for all individuals. 00:11:38.010 --> 00:11:39.660 align:middle line:84% But some of them will be different. 00:11:39.660 --> 00:11:42.360 align:middle line:84% And this is how SNPs are discovered, 00:11:42.360 --> 00:11:44.580 align:middle line:84% how single nucleotide polymorphisms are discovered. 00:11:44.580 --> 00:11:48.900 align:middle line:84% They are locations where different individuals don't 00:11:48.900 --> 00:11:50.760 align:middle line:90% have the exact same nucleotide. 00:11:50.760 --> 00:11:55.522 align:middle line:84% And we'll talk later about why genotyping is important, 00:11:55.522 --> 00:11:57.480 align:middle line:84% what kind of information you can get from that, 00:11:57.480 --> 00:12:02.850 align:middle line:84% and how this relates to diseases, to basically trying 00:12:02.850 --> 00:12:05.430 align:middle line:84% to figure out what is the relationship 00:12:05.430 --> 00:12:07.200 align:middle line:84% between the genotype, and in this case, 00:12:07.200 --> 00:12:11.893 align:middle line:84% between polymorphisms in your genotype and phenotype, 00:12:11.893 --> 00:12:12.435 align:middle line:90% like disease. 00:12:12.435 --> 00:12:15.580 align:middle line:90% 00:12:15.580 --> 00:12:21.250 align:middle line:84% At a level of RNA, it might be interesting to look 00:12:21.250 --> 00:12:23.500 align:middle line:84% at alternative splicing transcription rearrangements-- 00:12:23.500 --> 00:12:28.510 align:middle line:84% these are all things that happen to the original DNA sequence 00:12:28.510 --> 00:12:30.190 align:middle line:90% when it's transcribed into RNA. 00:12:30.190 --> 00:12:32.920 align:middle line:84% It undergoes a series of transformations 00:12:32.920 --> 00:12:37.540 align:middle line:84% that can, of course, affect, in a very 00:12:37.540 --> 00:12:42.140 align:middle line:90% deep way, the final product. 00:12:42.140 --> 00:12:48.720 align:middle line:84% And this process, the process of transcribing the DNA sequences 00:12:48.720 --> 00:12:54.540 align:middle line:84% into RNA, is, of course, at the basis of expression analysis. 00:12:54.540 --> 00:12:58.290 align:middle line:84% And you've heard a lot in the other lectures 00:12:58.290 --> 00:12:59.310 align:middle line:90% about microarrays. 00:12:59.310 --> 00:13:03.000 align:middle line:84% So I'm not going to go into too much details 00:13:03.000 --> 00:13:04.830 align:middle line:84% on this, but differentiable analysis, 00:13:04.830 --> 00:13:08.190 align:middle line:84% clustering, and so on, these are all the usual things that 00:13:08.190 --> 00:13:10.440 align:middle line:84% can be done using microarrays, using gene expression 00:13:10.440 --> 00:13:12.540 align:middle line:90% microarrays. 00:13:12.540 --> 00:13:13.890 align:middle line:90% We talk about proteins-- 00:13:13.890 --> 00:13:16.620 align:middle line:90% 00:13:16.620 --> 00:13:18.833 align:middle line:84% if you're studying a protein, the interesting thing 00:13:18.833 --> 00:13:20.250 align:middle line:84% is to do with a protein are trying 00:13:20.250 --> 00:13:25.440 align:middle line:84% to predict its active domains, in order to have an idea of how 00:13:25.440 --> 00:13:27.720 align:middle line:84% the protein might behave, what function it might have, 00:13:27.720 --> 00:13:30.750 align:middle line:84% how it might interact with other proteins, with other genes, 00:13:30.750 --> 00:13:32.430 align:middle line:90% and so on. 00:13:32.430 --> 00:13:34.890 align:middle line:84% Predicting the three dimensional structure of a protein 00:13:34.890 --> 00:13:42.340 align:middle line:84% is another very important and very complex task. 00:13:42.340 --> 00:13:43.930 align:middle line:84% Starting homology and conservation 00:13:43.930 --> 00:13:47.710 align:middle line:84% of proteins across different organisms 00:13:47.710 --> 00:13:56.650 align:middle line:84% can give you a very good idea of the importance 00:13:56.650 --> 00:13:57.820 align:middle line:90% of some proteins. 00:13:57.820 --> 00:14:03.340 align:middle line:84% So if something that's been around for millions of years, 00:14:03.340 --> 00:14:04.840 align:middle line:84% it probably means that it's involved 00:14:04.840 --> 00:14:06.730 align:middle line:84% in a very basic mechanism, while there 00:14:06.730 --> 00:14:09.640 align:middle line:84% are some proteins that are new, that are only 00:14:09.640 --> 00:14:10.720 align:middle line:90% specific to human beings. 00:14:10.720 --> 00:14:15.240 align:middle line:84% And that, again, can give you some information. 00:14:15.240 --> 00:14:18.080 align:middle line:84% And finally, something that is very challenging, 00:14:18.080 --> 00:14:20.600 align:middle line:84% and it's receiving a lot of attention lately, 00:14:20.600 --> 00:14:23.060 align:middle line:84% is the automatic construction and analysis 00:14:23.060 --> 00:14:26.390 align:middle line:84% of metabolic pathways and regulatory pathways. 00:14:26.390 --> 00:14:29.030 align:middle line:84% If you are able to understand how proteins interact 00:14:29.030 --> 00:14:34.100 align:middle line:84% with each other, and interact with the rest of the cell, how 00:14:34.100 --> 00:14:38.780 align:middle line:84% they then regulate other genes, and in turn other proteins, 00:14:38.780 --> 00:14:40.160 align:middle line:84% then you can use this information 00:14:40.160 --> 00:14:43.490 align:middle line:84% to try to build, in a computational way, the kind 00:14:43.490 --> 00:14:47.660 align:middle line:84% of pathway maps that biologists have been 00:14:47.660 --> 00:14:50.540 align:middle line:90% drawing by hand for decades. 00:14:50.540 --> 00:14:54.020 align:middle line:84% And of course, we're still very far away 00:14:54.020 --> 00:14:56.300 align:middle line:84% from being able to do this in the general case. 00:14:56.300 --> 00:14:59.280 align:middle line:90% 00:14:59.280 --> 00:15:01.400 align:middle line:84% It works in some limited case, and we're 00:15:01.400 --> 00:15:06.090 align:middle line:84% going to see later some examples of some of these things. 00:15:06.090 --> 00:15:08.150 align:middle line:84% But these are all very challenging 00:15:08.150 --> 00:15:13.310 align:middle line:84% problems that, of course, are still very much open. 00:15:13.310 --> 00:15:16.820 align:middle line:84% And finally, we get to the phenotype then, 00:15:16.820 --> 00:15:18.860 align:middle line:84% we could put a very long list of things here, 00:15:18.860 --> 00:15:21.140 align:middle line:84% but could talk about population genetics, 00:15:21.140 --> 00:15:23.060 align:middle line:90% about association studies. 00:15:23.060 --> 00:15:24.560 align:middle line:84% Association studies are studies that 00:15:24.560 --> 00:15:29.990 align:middle line:84% try to correlate the presence of a certain genotype 00:15:29.990 --> 00:15:31.460 align:middle line:90% with an observed phenotype. 00:15:31.460 --> 00:15:37.220 align:middle line:84% Like in the most common case, association studies based 00:15:37.220 --> 00:15:39.890 align:middle line:84% on SNPs, they just look at two different alleles of a SNP, 00:15:39.890 --> 00:15:42.140 align:middle line:84% and they try to figure out if there 00:15:42.140 --> 00:15:44.120 align:middle line:84% is statistical correlation between one 00:15:44.120 --> 00:15:46.280 align:middle line:84% or two alleles and the disease, and that 00:15:46.280 --> 00:15:47.840 align:middle line:84% might mean that the SNP is indeed 00:15:47.840 --> 00:15:53.150 align:middle line:84% responsible for the disease in some way. 00:15:53.150 --> 00:15:57.180 align:middle line:84% And clinical trials, of course, to validate all this. 00:15:57.180 --> 00:16:02.210 align:middle line:84% OK, so two more slides about philosophy 00:16:02.210 --> 00:16:10.430 align:middle line:84% and then we'll start with the more practical stuff. 00:16:10.430 --> 00:16:13.430 align:middle line:84% I've already mentioned the word gene a lot of times, 00:16:13.430 --> 00:16:17.090 align:middle line:84% and I'm going to mention it again very often. 00:16:17.090 --> 00:16:20.870 align:middle line:84% So it might be interesting to ask ourselves, what is a gene? 00:16:20.870 --> 00:16:25.190 align:middle line:84% And this is something that it's a question to which the answer 00:16:25.190 --> 00:16:28.647 align:middle line:84% is probably obvious, but it turns out 00:16:28.647 --> 00:16:30.980 align:middle line:84% that there are actually many possible answers, according 00:16:30.980 --> 00:16:34.640 align:middle line:84% to the context you're in, according to the different view 00:16:34.640 --> 00:16:38.880 align:middle line:90% of the world that you're using. 00:16:38.880 --> 00:16:41.953 align:middle line:84% So for example, if you ask a classical geneticist what 00:16:41.953 --> 00:16:43.370 align:middle line:84% is a gene, you will get the answer 00:16:43.370 --> 00:16:46.500 align:middle line:84% that a gene is the smallest unit of inheritance. 00:16:46.500 --> 00:16:52.190 align:middle line:84% This is the definition that goes back to Mendel, essentially. 00:16:52.190 --> 00:16:56.000 align:middle line:84% If you ask someone who is doing medical research, 00:16:56.000 --> 00:16:58.430 align:middle line:84% you will get the answer that the gene 00:16:58.430 --> 00:16:59.640 align:middle line:90% is a disease-causing trait. 00:16:59.640 --> 00:17:04.880 align:middle line:84% So hear about the [INAUDIBLE] gene, or the gene 00:17:04.880 --> 00:17:06.780 align:middle line:90% for cystic fibrosis, and so on. 00:17:06.780 --> 00:17:12.079 align:middle line:84% So in this case, the word gene has a very clear 00:17:12.079 --> 00:17:14.185 align:middle line:90% clinical connotation. 00:17:14.185 --> 00:17:15.560 align:middle line:84% If you ask a molecular biologist, 00:17:15.560 --> 00:17:17.810 align:middle line:84% you get the answer that the gene is a recipe-- 00:17:17.810 --> 00:17:22.760 align:middle line:84% is essentially a program to build one or more proteins. 00:17:22.760 --> 00:17:25.178 align:middle line:84% And we can go on, we can ask biochemists, 00:17:25.178 --> 00:17:27.470 align:middle line:84% and you will get the answer that the gene is an element 00:17:27.470 --> 00:17:28.970 align:middle line:90% in a metabolic network. 00:17:28.970 --> 00:17:34.400 align:middle line:84% It's an active element in one of those big networks 00:17:34.400 --> 00:17:39.230 align:middle line:84% of interacting genes that regulate each other, 00:17:39.230 --> 00:17:46.010 align:middle line:84% and that overall realize the metabolic process. 00:17:46.010 --> 00:17:47.720 align:middle line:84% If you ask a motor geneticist, you 00:17:47.720 --> 00:17:49.430 align:middle line:84% will get the answer that the gene 00:17:49.430 --> 00:17:52.490 align:middle line:84% is a locus on a chromosome, in a certain region 00:17:52.490 --> 00:17:56.720 align:middle line:84% of a chromosome, that has a functional characterization. 00:17:56.720 --> 00:17:59.150 align:middle line:84% Locus that was studied, and was found 00:17:59.150 --> 00:18:04.550 align:middle line:84% to have a specific function in our biology. 00:18:04.550 --> 00:18:06.450 align:middle line:84% And finally, if you ask a bioinformatician, 00:18:06.450 --> 00:18:10.520 align:middle line:84% you will get the answer that a gene is just a stretch of DNA 00:18:10.520 --> 00:18:13.905 align:middle line:84% where we know there is a gene, because the database tells us 00:18:13.905 --> 00:18:15.030 align:middle line:90% that there is a gene there. 00:18:15.030 --> 00:18:17.060 align:middle line:84% So we know it has a transcription start 00:18:17.060 --> 00:18:19.970 align:middle line:84% site, encoding sequence start site, has exons and intrinsic 00:18:19.970 --> 00:18:22.393 align:middle line:90% in certain positions. 00:18:22.393 --> 00:18:23.810 align:middle line:84% So in the following, we are to see 00:18:23.810 --> 00:18:26.780 align:middle line:84% examples of all of these different ways 00:18:26.780 --> 00:18:27.815 align:middle line:90% to look at the gene. 00:18:27.815 --> 00:18:32.540 align:middle line:90% 00:18:32.540 --> 00:18:34.040 align:middle line:84% And to start, of course, we're going 00:18:34.040 --> 00:18:35.910 align:middle line:90% to start from the beginning. 00:18:35.910 --> 00:18:37.350 align:middle line:90% So from DNA sequence data. 00:18:37.350 --> 00:18:41.572 align:middle line:84% And now, we're starting to look at where 00:18:41.572 --> 00:18:43.280 align:middle line:84% all these different pieces of information 00:18:43.280 --> 00:18:44.655 align:middle line:84% I've told you about can be found, 00:18:44.655 --> 00:18:49.080 align:middle line:84% and how they're stored and represented. 00:18:49.080 --> 00:18:51.320 align:middle line:84% So if we're talking about DNA sequence data, 00:18:51.320 --> 00:18:53.990 align:middle line:84% the first place to go is, of course, GenBank. 00:18:53.990 --> 00:18:58.340 align:middle line:84% GenBank is the largest repository of sequence data. 00:18:58.340 --> 00:19:00.810 align:middle line:84% It accepts direct submissions from researchers. 00:19:00.810 --> 00:19:03.770 align:middle line:84% So anybody in the world who sequences a new piece of DNA 00:19:03.770 --> 00:19:06.815 align:middle line:84% can send it to GenBank, and it's put in the big cauldron. 00:19:06.815 --> 00:19:10.340 align:middle line:90% 00:19:10.340 --> 00:19:12.200 align:middle line:90% This is data from-- 00:19:12.200 --> 00:19:14.840 align:middle line:84% the most recent data that I could find from one year ago. 00:19:14.840 --> 00:19:17.390 align:middle line:84% It contained more than 22 million sequences 00:19:17.390 --> 00:19:20.330 align:middle line:90% and 100,000 distinct organisms. 00:19:20.330 --> 00:19:23.210 align:middle line:84% With a total of almost 30 billion nucleotides. 00:19:23.210 --> 00:19:26.090 align:middle line:90% 00:19:26.090 --> 00:19:30.200 align:middle line:90% And this is the URL for GenBank. 00:19:30.200 --> 00:19:33.122 align:middle line:84% And GenBank is at the basis of the NCBI cluster. 00:19:33.122 --> 00:19:35.330 align:middle line:84% So the National Center for Biotechnology Information, 00:19:35.330 --> 00:19:40.040 align:middle line:84% is a branch of the NIH, that has the task of assembling 00:19:40.040 --> 00:19:42.740 align:middle line:84% the largest possible number of databases 00:19:42.740 --> 00:19:44.630 align:middle line:90% of biomedical information. 00:19:44.630 --> 00:19:48.920 align:middle line:84% They manage GenBank, and GenBank, in turn, 00:19:48.920 --> 00:19:51.290 align:middle line:84% is at the basis for a lot of other resources 00:19:51.290 --> 00:19:55.060 align:middle line:84% that we're going to see now, that 00:19:55.060 --> 00:19:58.945 align:middle line:84% are all part of this cluster of NCBI resources. 00:19:58.945 --> 00:20:00.820 align:middle line:84% They're all interconnected, so you can easily 00:20:00.820 --> 00:20:04.750 align:middle line:84% jump from one to the other, and that is a very powerful way 00:20:04.750 --> 00:20:09.370 align:middle line:90% of exploring this kind of data. 00:20:09.370 --> 00:20:11.790 align:middle line:84% So this is a graph that shows you the growth of GenBank 00:20:11.790 --> 00:20:12.600 align:middle line:90% in recent years. 00:20:12.600 --> 00:20:18.540 align:middle line:84% You can see the very steep growth 00:20:18.540 --> 00:20:19.870 align:middle line:90% of the number of base pairs. 00:20:19.870 --> 00:20:24.690 align:middle line:84% And the almost equally steep rate growth of the number of 00:20:24.690 --> 00:20:25.200 align:middle line:90% sequences. 00:20:25.200 --> 00:20:30.420 align:middle line:84% And you can probably tell that the number-- 00:20:30.420 --> 00:20:33.240 align:middle line:84% we're sequencing longer and longer sequences, 00:20:33.240 --> 00:20:37.830 align:middle line:84% because the blue graph grows more rapidly than the red one. 00:20:37.830 --> 00:20:43.110 align:middle line:84% But anyway, what do you do when you have all these accumulated 00:20:43.110 --> 00:20:43.660 align:middle line:90% sequences? 00:20:43.660 --> 00:20:48.180 align:middle line:84% So in GenBank, you just have sequences by themselves 00:20:48.180 --> 00:20:49.723 align:middle line:84% that can be very short, very long, 00:20:49.723 --> 00:20:52.140 align:middle line:84% but they're just independent sequences that were put there 00:20:52.140 --> 00:20:53.772 align:middle line:90% by investigators. 00:20:53.772 --> 00:20:55.980 align:middle line:84% So the thing you can do, if you have enough sequences 00:20:55.980 --> 00:20:58.045 align:middle line:84% from the same organism, you can try 00:20:58.045 --> 00:20:59.670 align:middle line:84% assembling them, putting them together, 00:20:59.670 --> 00:21:02.310 align:middle line:84% and trying to reconstruct the entire genome. 00:21:02.310 --> 00:21:07.170 align:middle line:84% And this is what was done to assemble 00:21:07.170 --> 00:21:09.870 align:middle line:84% the human genome, for example, and all the other genomes 00:21:09.870 --> 00:21:12.450 align:middle line:90% that are being sequenced. 00:21:12.450 --> 00:21:17.460 align:middle line:84% You start with-- you look at the sequences you have, 00:21:17.460 --> 00:21:19.980 align:middle line:84% and if you can find overlaps, then 00:21:19.980 --> 00:21:22.590 align:middle line:84% you know that these two sequences are 00:21:22.590 --> 00:21:24.970 align:middle line:84% related in some way, and you proceed from there. 00:21:24.970 --> 00:21:28.350 align:middle line:84% So let's say you've sequenced this sequence, then 00:21:28.350 --> 00:21:33.660 align:middle line:84% another one, they are distinct and they have an overlap, 00:21:33.660 --> 00:21:35.640 align:middle line:84% so you don't know where they are on the genome. 00:21:35.640 --> 00:21:38.550 align:middle line:84% But then, if you sequence a third one that 00:21:38.550 --> 00:21:41.250 align:middle line:84% overlaps the first one and includes the second one, 00:21:41.250 --> 00:21:43.150 align:middle line:84% then you know that you can basically 00:21:43.150 --> 00:21:45.720 align:middle line:90% arrange them in this way. 00:21:45.720 --> 00:21:48.520 align:middle line:84% And then let's say you sequence another one, 00:21:48.520 --> 00:21:51.120 align:middle line:84% and again you find an overlap with one that you already have. 00:21:51.120 --> 00:21:52.590 align:middle line:84% So like this, and in the end you're 00:21:52.590 --> 00:21:56.850 align:middle line:84% going to build a map that tells you 00:21:56.850 --> 00:21:59.070 align:middle line:84% where all these fragments should be 00:21:59.070 --> 00:22:00.990 align:middle line:90% positioned on the chromosome. 00:22:00.990 --> 00:22:03.010 align:middle line:84% And you have different levels of coverage, 00:22:03.010 --> 00:22:05.400 align:middle line:84% so you have regions that you've seen only once, regions 00:22:05.400 --> 00:22:08.587 align:middle line:84% you've seen twice, regions you've seen three times. 00:22:08.587 --> 00:22:11.170 align:middle line:84% If you have a sufficient degree of coverage, then you can say, 00:22:11.170 --> 00:22:15.090 align:middle line:84% well, OK, I believe this is the right sequence. 00:22:15.090 --> 00:22:18.700 align:middle line:84% And you proceed from there, you take that for granted, 00:22:18.700 --> 00:22:20.430 align:middle line:84% you proceed from there by connecting more 00:22:20.430 --> 00:22:22.255 align:middle line:90% and more overlapping pieces. 00:22:22.255 --> 00:22:23.880 align:middle line:84% And again, this is how the human genome 00:22:23.880 --> 00:22:27.660 align:middle line:90% was sequenced essentially. 00:22:27.660 --> 00:22:32.110 align:middle line:90% Up to a level of coverage-- 00:22:32.110 --> 00:22:33.840 align:middle line:84% I think it's five or six times covered, 00:22:33.840 --> 00:22:35.850 align:middle line:84% so each stretch of DNA in the human genome 00:22:35.850 --> 00:22:43.120 align:middle line:84% has been sequenced at least five or six times for validation. 00:22:43.120 --> 00:22:47.880 align:middle line:84% So and the details of how this process has been implemented 00:22:47.880 --> 00:22:48.825 align:middle line:90% and CVI are here. 00:22:48.825 --> 00:22:53.412 align:middle line:90% 00:22:53.412 --> 00:22:55.370 align:middle line:84% And in addition to the human genome, of course, 00:22:55.370 --> 00:22:57.453 align:middle line:84% we have a lot of other genomes that are completed, 00:22:57.453 --> 00:22:59.500 align:middle line:90% or near completion. 00:22:59.500 --> 00:23:02.020 align:middle line:84% These numbers are probably higher by now 00:23:02.020 --> 00:23:03.580 align:middle line:90% we have over 1,000 viruses. 00:23:03.580 --> 00:23:09.460 align:middle line:84% And then many other organisms from different domains of life. 00:23:09.460 --> 00:23:13.420 align:middle line:84% Of course, eukaryotes are the hardest organisms to sequence. 00:23:13.420 --> 00:23:18.370 align:middle line:84% But the human genome is considered finished by now. 00:23:18.370 --> 00:23:25.380 align:middle line:84% It's hard to go above this level of accuracy. 00:23:25.380 --> 00:23:29.230 align:middle line:84% It's probably not even necessary because the differences 00:23:29.230 --> 00:23:31.570 align:middle line:84% between two human beings are probably 00:23:31.570 --> 00:23:35.200 align:middle line:90% the same order of magnitude. 00:23:35.200 --> 00:23:38.710 align:middle line:84% But now we have several other organisms, including a chimp, 00:23:38.710 --> 00:23:39.970 align:middle line:90% that was recently released. 00:23:39.970 --> 00:23:41.680 align:middle line:84% And it's going to be very interesting 00:23:41.680 --> 00:23:45.680 align:middle line:84% because it's essentially identical to the human genome. 00:23:45.680 --> 00:23:48.430 align:middle line:90% So there is 1%-- 00:23:48.430 --> 00:23:50.875 align:middle line:84% the differences are about 1% between the genomes 00:23:50.875 --> 00:23:52.000 align:middle line:90% of the human and the chimp. 00:23:52.000 --> 00:23:53.110 align:middle line:84% And it will be very interesting to see 00:23:53.110 --> 00:23:54.640 align:middle line:84% exactly where these differences are, 00:23:54.640 --> 00:23:58.780 align:middle line:84% and why these differences are responsible for the fact 00:23:58.780 --> 00:24:02.950 align:middle line:84% that we are humans and they are monkeys. 00:24:02.950 --> 00:24:08.290 align:middle line:84% Again this is a link to the entry point 00:24:08.290 --> 00:24:09.940 align:middle line:84% for the description of all the genomes 00:24:09.940 --> 00:24:13.240 align:middle line:90% that are stored in NCBI. 00:24:13.240 --> 00:24:16.060 align:middle line:84% And a new genomes, the small ones-- the viruses 00:24:16.060 --> 00:24:18.620 align:middle line:84% and bacteria, there are new genomes, essentially, 00:24:18.620 --> 00:24:19.120 align:middle line:90% every week. 00:24:19.120 --> 00:24:21.535 align:middle line:84% So these numbers constantly change. 00:24:21.535 --> 00:24:24.090 align:middle line:90% 00:24:24.090 --> 00:24:28.230 align:middle line:84% Now, we now have the complete subsequence 00:24:28.230 --> 00:24:29.100 align:middle line:90% of the human genome. 00:24:29.100 --> 00:24:30.610 align:middle line:90% Where do you find it? 00:24:30.610 --> 00:24:32.880 align:middle line:84% So the best resource, in my opinion, 00:24:32.880 --> 00:24:40.170 align:middle line:84% for looking at the human genome in literally is Golden Path. 00:24:40.170 --> 00:24:43.680 align:middle line:84% Golden Path is a genome browser for 00:24:43.680 --> 00:24:44.945 align:middle line:90% several different organisms. 00:24:44.945 --> 00:24:46.320 align:middle line:84% Initially it was only for human-- 00:24:46.320 --> 00:24:50.310 align:middle line:84% now it has mouse, rat, chimp, Drosophila, yeast, 00:24:50.310 --> 00:24:52.930 align:middle line:90% and a few others. 00:24:52.930 --> 00:24:54.820 align:middle line:84% The nice thing about Golden Path is 00:24:54.820 --> 00:24:58.990 align:middle line:84% that it gives a graphical view, so you 00:24:58.990 --> 00:25:01.810 align:middle line:84% can see it in the next slide, it's very clear, 00:25:01.810 --> 00:25:04.510 align:middle line:84% it's very easy to find all the information 00:25:04.510 --> 00:25:07.480 align:middle line:84% you need about a certain region of the chromosome. 00:25:07.480 --> 00:25:10.420 align:middle line:84% On the other hand, all the information it provides 00:25:10.420 --> 00:25:14.890 align:middle line:84% is available in easy to download and to parse formats. 00:25:14.890 --> 00:25:16.640 align:middle line:84% So if you want to build your own database, 00:25:16.640 --> 00:25:18.370 align:middle line:84% it contains the same information that's 00:25:18.370 --> 00:25:21.200 align:middle line:90% something that is pretty easy. 00:25:21.200 --> 00:25:23.440 align:middle line:84% It provides arbitrary DNA sequences-- 00:25:23.440 --> 00:25:26.980 align:middle line:84% so you can ask for any region of any human chromosome, 00:25:26.980 --> 00:25:29.650 align:middle line:84% you'll get back the exact DNA sequence for that region. 00:25:29.650 --> 00:25:36.700 align:middle line:84% For something that might think it's easy by now, 00:25:36.700 --> 00:25:39.820 align:middle line:84% but two or three years ago, it was still 00:25:39.820 --> 00:25:43.570 align:middle line:84% something that was almost impossible to obtain. 00:25:43.570 --> 00:25:45.100 align:middle line:84% And Golden Path was the first site 00:25:45.100 --> 00:25:47.410 align:middle line:90% to provide something like this. 00:25:47.410 --> 00:25:51.262 align:middle line:84% It gives you the absolute position of all the known 00:25:51.262 --> 00:25:52.220 align:middle line:90% elements of our genome. 00:25:52.220 --> 00:25:54.400 align:middle line:84% So genes, markers, mutations, other features, 00:25:54.400 --> 00:25:56.500 align:middle line:84% they tell you exactly they are at this location, 00:25:56.500 --> 00:25:59.260 align:middle line:84% in terms of the absolute base pair. 00:25:59.260 --> 00:26:01.810 align:middle line:90% 00:26:01.810 --> 00:26:04.720 align:middle line:90% This is the URL for Golden Path. 00:26:04.720 --> 00:26:07.670 align:middle line:84% And this is how it looks like-- this is an example, 00:26:07.670 --> 00:26:11.350 align:middle line:84% we're looking at the region that contains the TLR1 gene. 00:26:11.350 --> 00:26:14.320 align:middle line:84% And you can see all these different tracks 00:26:14.320 --> 00:26:18.370 align:middle line:84% that provide different information 00:26:18.370 --> 00:26:19.310 align:middle line:90% on different objects. 00:26:19.310 --> 00:26:21.060 align:middle line:84% So for example, up here we have the genes, 00:26:21.060 --> 00:26:23.500 align:middle line:84% we have different sets of known genes. 00:26:23.500 --> 00:26:24.970 align:middle line:84% We have predicted genes, according 00:26:24.970 --> 00:26:27.820 align:middle line:84% to different prediction algorithms. 00:26:27.820 --> 00:26:31.870 align:middle line:84% We have the mRNAs that were aligned to this gene. 00:26:31.870 --> 00:26:34.540 align:middle line:90% 00:26:34.540 --> 00:26:36.510 align:middle line:84% We have a tract like this and that 00:26:36.510 --> 00:26:39.430 align:middle line:84% shows the conservation between human and mouse in this region. 00:26:39.430 --> 00:26:41.190 align:middle line:84% So you can see it's very interesting 00:26:41.190 --> 00:26:45.570 align:middle line:84% to see that obviously the coding part of the gene 00:26:45.570 --> 00:26:48.173 align:middle line:84% is the one that is most highly conserved between human 00:26:48.173 --> 00:26:49.840 align:middle line:84% and mouse-- that is the functional part. 00:26:49.840 --> 00:26:52.785 align:middle line:84% So it undergoes selective pressure. 00:26:52.785 --> 00:26:55.373 align:middle line:90% 00:26:55.373 --> 00:26:57.540 align:middle line:84% There are tracks that tell you the location of SNPs, 00:26:57.540 --> 00:26:58.060 align:middle line:90% and so on. 00:26:58.060 --> 00:27:00.090 align:middle line:84% There are many others that of course don't fit in here, 00:27:00.090 --> 00:27:01.440 align:middle line:84% but you can customize the display. 00:27:01.440 --> 00:27:03.190 align:middle line:84% You can select the tracks you want to see, 00:27:03.190 --> 00:27:07.860 align:middle line:84% and you get your own view of a certain genetic region. 00:27:07.860 --> 00:27:12.750 align:middle line:84% You have the coordinates up here, chromosome 4, the band, 00:27:12.750 --> 00:27:15.880 align:middle line:90% and so on. 00:27:15.880 --> 00:27:19.170 align:middle line:84% And this is just to show that you can query it 00:27:19.170 --> 00:27:23.232 align:middle line:84% for any-- this is the same DNA region we 00:27:23.232 --> 00:27:24.190 align:middle line:90% were looking at before. 00:27:24.190 --> 00:27:25.860 align:middle line:84% But in this case, we asked for the DNA sequence, 00:27:25.860 --> 00:27:26.520 align:middle line:90% and we get it. 00:27:26.520 --> 00:27:29.680 align:middle line:90% 00:27:29.680 --> 00:27:32.650 align:middle line:84% There's another way of looking at the genome using the NCBI 00:27:32.650 --> 00:27:33.630 align:middle line:90% map viewer. 00:27:33.630 --> 00:27:35.230 align:middle line:84% It's essentially the same thing-- 00:27:35.230 --> 00:27:38.370 align:middle line:84% it's a graphical browser to look at genomes 00:27:38.370 --> 00:27:42.030 align:middle line:90% and annotations of genomes. 00:27:42.030 --> 00:27:43.860 align:middle line:84% It's organized around several maps-- 00:27:43.860 --> 00:27:47.340 align:middle line:84% they have sequence maps, cytogenetic maps, 00:27:47.340 --> 00:27:50.790 align:middle line:84% language maps, radiation hybrid, human mouse homology maps. 00:27:50.790 --> 00:27:52.260 align:middle line:84% So each one of these is a view that 00:27:52.260 --> 00:27:58.940 align:middle line:84% gives you a different set of objects in the view. 00:27:58.940 --> 00:28:00.940 align:middle line:84% So in the sequence map, you can find information 00:28:00.940 --> 00:28:05.410 align:middle line:84% about the genes, transcripts, gene clusters, and so on. 00:28:05.410 --> 00:28:06.910 align:middle line:84% Inside the genetic map, for example, 00:28:06.910 --> 00:28:10.630 align:middle line:84% you find information about disease genes, 00:28:10.630 --> 00:28:13.720 align:middle line:90% what bands break points. 00:28:13.720 --> 00:28:15.960 align:middle line:84% It's extremely detailed, because of course, it 00:28:15.960 --> 00:28:19.030 align:middle line:84% can rely on the whole set of NCBI databases, 00:28:19.030 --> 00:28:25.480 align:middle line:84% so basically everything that could possibly want to know 00:28:25.480 --> 00:28:28.010 align:middle line:90% is in there. 00:28:28.010 --> 00:28:31.570 align:middle line:84% I, personally, find it a bit complex to use, a bit harder 00:28:31.570 --> 00:28:33.380 align:middle line:90% to use than Golden Path. 00:28:33.380 --> 00:28:34.660 align:middle line:90% But it's is a matter of taste. 00:28:34.660 --> 00:28:37.990 align:middle line:84% It's organized in a different way. 00:28:37.990 --> 00:28:39.970 align:middle line:84% Shows you much more detailed information 00:28:39.970 --> 00:28:42.580 align:middle line:84% here in this graphical bars, on the features 00:28:42.580 --> 00:28:44.860 align:middle line:90% of the genetic region. 00:28:44.860 --> 00:28:47.267 align:middle line:84% But then the coded information is a bit harder. 00:28:47.267 --> 00:28:48.850 align:middle line:84% They use all these abbreviations here. 00:28:48.850 --> 00:28:51.490 align:middle line:90% So it's a matter of taste. 00:28:51.490 --> 00:28:53.860 align:middle line:84% They essentially serve the same purpose 00:28:53.860 --> 00:28:59.447 align:middle line:84% with different levels of detail in different areas. 00:28:59.447 --> 00:29:00.655 align:middle line:90% Now, we've talked about SNPs. 00:29:00.655 --> 00:29:04.975 align:middle line:90% 00:29:04.975 --> 00:29:07.350 align:middle line:84% It's the only form of mutation we're going to talk about, 00:29:07.350 --> 00:29:09.150 align:middle line:84% but it's also the most important one. 00:29:09.150 --> 00:29:12.300 align:middle line:84% Because first of all, SNPs are the most common form 00:29:12.300 --> 00:29:13.425 align:middle line:90% of variation in our genome. 00:29:13.425 --> 00:29:15.960 align:middle line:90% 00:29:15.960 --> 00:29:19.150 align:middle line:84% They're much more frequent than microsatellites or insertions, 00:29:19.150 --> 00:29:20.790 align:middle line:90% deletions, and other things. 00:29:20.790 --> 00:29:22.950 align:middle line:84% And they're important because for example, they 00:29:22.950 --> 00:29:24.750 align:middle line:90% can be used as genomic markers. 00:29:24.750 --> 00:29:28.050 align:middle line:84% So SNPs are at a fixed location in the genome. 00:29:28.050 --> 00:29:30.690 align:middle line:84% And if you know where the SNP is, you can find it-- 00:29:30.690 --> 00:29:36.118 align:middle line:84% you can find the same location in different individuals. 00:29:36.118 --> 00:29:37.410 align:middle line:90% So you can use them as markers. 00:29:37.410 --> 00:29:39.535 align:middle line:84% You can use them as causal candidates for diseases, 00:29:39.535 --> 00:29:42.300 align:middle line:84% because a certain percentage of the SNPs 00:29:42.300 --> 00:29:50.400 align:middle line:84% introduce changes that then have some consequence 00:29:50.400 --> 00:29:54.855 align:middle line:84% on the genotype, on the phenotype. 00:29:54.855 --> 00:29:56.230 align:middle line:84% What I mean is that, for example, 00:29:56.230 --> 00:29:58.605 align:middle line:84% if you have a SNP in the coding subsequence of a protein, 00:29:58.605 --> 00:30:02.430 align:middle line:84% you're going to get a protein that has an abnormal sequence. 00:30:02.430 --> 00:30:06.060 align:middle line:84% And that can be a change that doesn't cause any consequence, 00:30:06.060 --> 00:30:08.610 align:middle line:84% or it might be a very dramatic change. 00:30:08.610 --> 00:30:13.680 align:middle line:84% For example, the most extreme example 00:30:13.680 --> 00:30:16.520 align:middle line:84% is there are some SNPs that introduce a stop in the protein 00:30:16.520 --> 00:30:17.020 align:middle line:90% sequence. 00:30:17.020 --> 00:30:19.200 align:middle line:84% So the protein sequence is truncated-- instead of just 00:30:19.200 --> 00:30:20.492 align:middle line:90% being modified, it's truncated. 00:30:20.492 --> 00:30:23.580 align:middle line:90% It's shorter than it should be. 00:30:23.580 --> 00:30:25.350 align:middle line:84% As you can imagine, this is a chain 00:30:25.350 --> 00:30:28.300 align:middle line:90% that can be very dangerous. 00:30:28.300 --> 00:30:30.300 align:middle line:84% There are many diseases that are due to the fact 00:30:30.300 --> 00:30:33.610 align:middle line:84% that you have SNPs that truncate proteins. 00:30:33.610 --> 00:30:35.610 align:middle line:84% They can be used as evolutionary markers, 00:30:35.610 --> 00:30:39.778 align:middle line:84% because SNPs arise randomly during the replication, 00:30:39.778 --> 00:30:41.820 align:middle line:84% and then they are transmitted from one generation 00:30:41.820 --> 00:30:43.530 align:middle line:90% to the next. 00:30:43.530 --> 00:30:52.630 align:middle line:84% And it's very interesting to study how SNPs get-- 00:30:52.630 --> 00:30:54.920 align:middle line:84% how the frequency of the SNP changes in a population. 00:30:54.920 --> 00:30:58.690 align:middle line:84% So if you have a SNP that provides an advantage for you-- 00:30:58.690 --> 00:31:00.730 align:middle line:84% because I mean, most SNPs are deleterious. 00:31:00.730 --> 00:31:02.350 align:middle line:84% But in some cases, the SNP can also 00:31:02.350 --> 00:31:04.480 align:middle line:84% provide an advantage, if it generates something 00:31:04.480 --> 00:31:07.090 align:middle line:84% that was not present before, and that 00:31:07.090 --> 00:31:08.810 align:middle line:90% works better than the original. 00:31:08.810 --> 00:31:13.240 align:middle line:84% So if you have a SNP that introduces a change is 00:31:13.240 --> 00:31:16.360 align:middle line:90% beneficial, then you will-- 00:31:16.360 --> 00:31:19.090 align:middle line:84% given enough time-- you will see that the frequency of the SNPs 00:31:19.090 --> 00:31:21.340 align:middle line:84% increases in the population, more and more individuals 00:31:21.340 --> 00:31:25.630 align:middle line:84% are going to have the variant form of the SNP. 00:31:25.630 --> 00:31:27.280 align:middle line:84% On the other hand, if a SNP is neutral, 00:31:27.280 --> 00:31:28.780 align:middle line:84% then there is no selective pressure, 00:31:28.780 --> 00:31:30.730 align:middle line:84% and it will either go away by chance, 00:31:30.730 --> 00:31:34.992 align:middle line:84% or will stay at a certain basic level of frequency. 00:31:34.992 --> 00:31:36.700 align:middle line:84% So you can study the frequency of the SNP 00:31:36.700 --> 00:31:39.790 align:middle line:84% to understand if it's undergoing selective pressure, so to know 00:31:39.790 --> 00:31:43.750 align:middle line:84% if it's deleterious or not, or you 00:31:43.750 --> 00:31:48.400 align:middle line:84% can use it to reconstruct basically 00:31:48.400 --> 00:31:50.528 align:middle line:90% the history of our genome. 00:31:50.528 --> 00:31:52.570 align:middle line:84% There are ways of calculating the age of the SNP, 00:31:52.570 --> 00:32:00.860 align:middle line:84% so when that mutation arose in the history of our genome. 00:32:00.860 --> 00:32:05.410 align:middle line:84% Now the largest database of SNPs that we have again, is at NCBI, 00:32:05.410 --> 00:32:08.230 align:middle line:90% it's called the dbSNP-- 00:32:08.230 --> 00:32:11.440 align:middle line:84% currently contains over 4 million human SNPs-- actually, 00:32:11.440 --> 00:32:14.920 align:middle line:84% I think that by now this number is closer to 5 million SNPs. 00:32:14.920 --> 00:32:17.650 align:middle line:84% And almost 50% of the SNPs are validated, 00:32:17.650 --> 00:32:19.550 align:middle line:84% which is something very important, 00:32:19.550 --> 00:32:22.930 align:middle line:84% means that the SNP has been observed independently 00:32:22.930 --> 00:32:25.660 align:middle line:90% multiple times. 00:32:25.660 --> 00:32:26.830 align:middle line:90% So you know it's a true SNP. 00:32:26.830 --> 00:32:27.330 align:middle line:90% 00:32:27.330 --> 00:32:30.850 align:middle line:84% It could be, many times, since sequencing is not 00:32:30.850 --> 00:32:34.263 align:middle line:84% an exact process, if you just look at a set of sequencing 00:32:34.263 --> 00:32:35.680 align:middle line:84% traces, you could think that there 00:32:35.680 --> 00:32:39.400 align:middle line:84% is a SNP when it's actually just a sequencing error. 00:32:39.400 --> 00:32:40.900 align:middle line:84% Now, if the SNP was validated, means 00:32:40.900 --> 00:32:44.260 align:middle line:84% it was observed several times by independent investigators, 00:32:44.260 --> 00:32:47.380 align:middle line:84% and that gives you the almost total certainty 00:32:47.380 --> 00:32:48.340 align:middle line:90% that it's a true SNP. 00:32:48.340 --> 00:32:51.640 align:middle line:90% 00:32:51.640 --> 00:32:54.790 align:middle line:84% There are other databases of SNPs. 00:32:54.790 --> 00:32:57.220 align:middle line:84% Another very important one is the SNP Consortium Database 00:32:57.220 --> 00:33:01.000 align:middle line:84% at Cold Spring Harbor, that offers-- 00:33:01.000 --> 00:33:04.810 align:middle line:84% the important thing about TSC is that first of all, all the TSC 00:33:04.810 --> 00:33:05.860 align:middle line:90% SNPs are validated. 00:33:05.860 --> 00:33:08.003 align:middle line:84% So they basically take SNPs on dbSNPs, 00:33:08.003 --> 00:33:09.670 align:middle line:84% then they check them again, to make sure 00:33:09.670 --> 00:33:10.930 align:middle line:90% that they're really SNPs. 00:33:10.930 --> 00:33:12.580 align:middle line:84% And while doing that, they also look 00:33:12.580 --> 00:33:14.650 align:middle line:90% at the frequency of the SNPs. 00:33:14.650 --> 00:33:15.430 align:middle line:90% So what do I mean? 00:33:15.430 --> 00:33:22.000 align:middle line:84% A SNP is a polymorphism that substitutes 00:33:22.000 --> 00:33:24.040 align:middle line:84% the nucleotide you should have at one 00:33:24.040 --> 00:33:26.080 align:middle line:90% location with a different one. 00:33:26.080 --> 00:33:31.360 align:middle line:84% So if you look at a population of individuals, 00:33:31.360 --> 00:33:34.660 align:middle line:84% you're going to see that the major allele of the SNP, 00:33:34.660 --> 00:33:37.490 align:middle line:84% the common one has a certain frequency, 00:33:37.490 --> 00:33:40.030 align:middle line:84% so it appears, for example, in 80% of individuals. 00:33:40.030 --> 00:33:42.460 align:middle line:84% And the alternative allele appears 00:33:42.460 --> 00:33:45.370 align:middle line:90% in 20% of the population. 00:33:45.370 --> 00:33:47.620 align:middle line:84% Now, knowing this frequency is very important, 00:33:47.620 --> 00:33:51.670 align:middle line:84% because it allows you, then, to do association studies. 00:33:51.670 --> 00:33:54.910 align:middle line:84% For example, to look for a correlation between a disease 00:33:54.910 --> 00:33:56.020 align:middle line:90% and this polymorphism. 00:33:56.020 --> 00:33:59.655 align:middle line:84% Because if then you observe a second population that 00:33:59.655 --> 00:34:01.030 align:middle line:84% is affected by a disease, and you 00:34:01.030 --> 00:34:02.920 align:middle line:84% find that in that second population, 00:34:02.920 --> 00:34:06.520 align:middle line:84% the alternative allele occurs with a frequency of 40% instead 00:34:06.520 --> 00:34:08.830 align:middle line:84% of 20, then that might be an indication 00:34:08.830 --> 00:34:10.940 align:middle line:84% that the SNP has something to do with the disease. 00:34:10.940 --> 00:34:12.260 align:middle line:84% But in order to be able to do this, 00:34:12.260 --> 00:34:14.260 align:middle line:84% you have to know what is the baseline frequency. 00:34:14.260 --> 00:34:19.210 align:middle line:84% What is the original frequency in normal, so to speak, 00:34:19.210 --> 00:34:21.800 align:middle line:90% human beings. 00:34:21.800 --> 00:34:23.090 align:middle line:90% Yes? 00:34:23.090 --> 00:34:26.040 align:middle line:84% AUDIENCE: I guess the question then becomes, 00:34:26.040 --> 00:34:29.437 align:middle line:84% what is the base population is across base population-- 00:34:29.437 --> 00:34:31.520 align:middle line:84% ALBERTO RIVA: That's what I was going to say next. 00:34:31.520 --> 00:34:33.020 align:middle line:84% Of course, the biggest problem here 00:34:33.020 --> 00:34:35.420 align:middle line:84% is that different populations may have 00:34:35.420 --> 00:34:37.550 align:middle line:90% different frequencies of SNPs. 00:34:37.550 --> 00:34:40.489 align:middle line:84% And this is one of the reasons why 00:34:40.489 --> 00:34:42.980 align:middle line:84% SNPs are used for population genetics, 00:34:42.980 --> 00:34:47.650 align:middle line:84% because especially in the past, when populations were 00:34:47.650 --> 00:34:49.810 align:middle line:90% much more closer there now. 00:34:49.810 --> 00:34:52.870 align:middle line:84% If a SNP arises in a population, then it 00:34:52.870 --> 00:34:54.969 align:middle line:84% tends to be limited to that population. 00:34:54.969 --> 00:34:58.060 align:middle line:84% You're not going to find it in a different population, 00:34:58.060 --> 00:35:01.578 align:middle line:84% unless there is some genetic interchange between the two. 00:35:01.578 --> 00:35:03.370 align:middle line:84% So when you look at the frequency of a SNP, 00:35:03.370 --> 00:35:05.950 align:middle line:84% it's very important to specify what population you're 00:35:05.950 --> 00:35:07.810 align:middle line:84% looking at, because we're going to have 00:35:07.810 --> 00:35:09.170 align:middle line:90% an example in two slides. 00:35:09.170 --> 00:35:12.370 align:middle line:84% So let's just get back to this in a second. 00:35:12.370 --> 00:35:14.620 align:middle line:84% Because I wanted to tell you about other SNP resources 00:35:14.620 --> 00:35:15.120 align:middle line:90% quickly. 00:35:15.120 --> 00:35:17.530 align:middle line:90% 00:35:17.530 --> 00:35:19.960 align:middle line:84% Haplotype Map project-- this is kind 00:35:19.960 --> 00:35:26.877 align:middle line:84% of a new project that is aimed at developing a haplotype map 00:35:26.877 --> 00:35:27.710 align:middle line:90% of the human genome. 00:35:27.710 --> 00:35:29.880 align:middle line:84% I don't if you had a lecture about haplotypes, 00:35:29.880 --> 00:35:31.830 align:middle line:90% about selecting-- 00:35:31.830 --> 00:35:33.540 align:middle line:84% you're going to-- OK, so you're going 00:35:33.540 --> 00:35:35.520 align:middle line:90% to have it later than this one. 00:35:35.520 --> 00:35:40.500 align:middle line:84% But when you hear about haplotypes, 00:35:40.500 --> 00:35:43.290 align:middle line:84% just remember that HapMap is a project that 00:35:43.290 --> 00:35:46.050 align:middle line:84% is aiming at building a complete haplotype map 00:35:46.050 --> 00:35:47.310 align:middle line:90% of the human genome. 00:35:47.310 --> 00:35:50.890 align:middle line:84% And don't have time to go into that now, 00:35:50.890 --> 00:35:52.710 align:middle line:84% but it's a very important resource 00:35:52.710 --> 00:35:56.490 align:middle line:84% that is really the next step after what TSC is doing, 00:35:56.490 --> 00:35:59.970 align:middle line:84% after determining the frequency of SNPs 00:35:59.970 --> 00:36:03.570 align:middle line:84% in different populations, that the HapMap project allows 00:36:03.570 --> 00:36:09.570 align:middle line:84% you to understand exactly what this can tell you 00:36:09.570 --> 00:36:12.150 align:middle line:84% about the evolution of our genome. 00:36:12.150 --> 00:36:17.420 align:middle line:84% But this will become clearer in the lecture about haplotypes. 00:36:17.420 --> 00:36:20.730 align:middle line:84% HGbase, another database of SNPs, 00:36:20.730 --> 00:36:23.060 align:middle line:84% it's manually curated, so you find-- 00:36:23.060 --> 00:36:25.320 align:middle line:84% it's very limited, but you find information 00:36:25.320 --> 00:36:27.340 align:middle line:90% that has very high quality. 00:36:27.340 --> 00:36:30.450 align:middle line:84% It's all manually verified, and it 00:36:30.450 --> 00:36:34.497 align:middle line:84% focuses on the potential consequences of SNPs. 00:36:34.497 --> 00:36:36.330 align:middle line:84% So you're going to find a lot of information 00:36:36.330 --> 00:36:40.950 align:middle line:84% about known associations between SNPs and diseases. 00:36:40.950 --> 00:36:44.400 align:middle line:84% Alfred at Yale is another very small database, 00:36:44.400 --> 00:36:47.310 align:middle line:84% but it has a very high quality, and it 00:36:47.310 --> 00:36:48.630 align:middle line:90% focuses on frequency data. 00:36:48.630 --> 00:36:51.240 align:middle line:84% And what they do is very, very interesting. 00:36:51.240 --> 00:36:55.500 align:middle line:84% They go look at many, many different populations, 00:36:55.500 --> 00:36:58.890 align:middle line:84% and especially a population, at small isolated populations 00:36:58.890 --> 00:37:01.530 align:middle line:84% from where we are, places like small islands, 00:37:01.530 --> 00:37:06.220 align:middle line:84% and Pacific or remote villages in Siberia, and so on. 00:37:06.220 --> 00:37:09.900 align:middle line:84% So they actually try to look for isolated populations 00:37:09.900 --> 00:37:14.880 align:middle line:84% to maximize the differences in SNP frequency 00:37:14.880 --> 00:37:16.380 align:middle line:84% that they're going to find, in order 00:37:16.380 --> 00:37:20.945 align:middle line:84% to have a picture of human diversity 00:37:20.945 --> 00:37:21.945 align:middle line:90% as complete as possible. 00:37:21.945 --> 00:37:24.820 align:middle line:90% 00:37:24.820 --> 00:37:28.500 align:middle line:84% And finally, SNPper that I'm citing because we developed 00:37:28.500 --> 00:37:32.460 align:middle line:84% a chip, this is a resource that tries 00:37:32.460 --> 00:37:35.070 align:middle line:84% to integrate information from all the places 00:37:35.070 --> 00:37:36.090 align:middle line:90% that I sited so far. 00:37:36.090 --> 00:37:39.210 align:middle line:84% So it takes information mainly from dbSNP, from Golden Path, 00:37:39.210 --> 00:37:44.910 align:middle line:84% from TSC, from Alfred, from HGbase, 00:37:44.910 --> 00:37:46.590 align:middle line:84% and it tries to put everything together 00:37:46.590 --> 00:37:50.280 align:middle line:84% in a unified view that allows you to look at the gene, 00:37:50.280 --> 00:37:52.170 align:middle line:84% find all the SNPs around that gene, 00:37:52.170 --> 00:37:53.980 align:middle line:84% see all the features of the SNPs, 00:37:53.980 --> 00:37:56.490 align:middle line:84% whether they are different decoding 00:37:56.490 --> 00:37:59.200 align:middle line:84% sequence, or the promoter sequence, or whatever. 00:37:59.200 --> 00:38:04.620 align:middle line:84% And then look at everything that is known about individual SNPs. 00:38:04.620 --> 00:38:08.640 align:middle line:84% And it provides a way of exporting this data 00:38:08.640 --> 00:38:11.220 align:middle line:84% in different formats to make it easier to process later. 00:38:11.220 --> 00:38:14.460 align:middle line:84% And I just want to show you one slide from SNPper, 00:38:14.460 --> 00:38:20.768 align:middle line:84% but this is a window that describes-- that tells you 00:38:20.768 --> 00:38:22.560 align:middle line:84% information about the particular SNP-- this 00:38:22.560 --> 00:38:24.780 align:middle line:90% is a [? SNP ?] identifier. 00:38:24.780 --> 00:38:26.970 align:middle line:84% And so you can see there is a top part where 00:38:26.970 --> 00:38:29.565 align:middle line:84% you have general information where the SNP is, just 00:38:29.565 --> 00:38:35.490 align:middle line:84% the position on chromosome 6, where the alleles are, 00:38:35.490 --> 00:38:38.310 align:middle line:90% the gene it belongs to, notch 4. 00:38:38.310 --> 00:38:40.440 align:middle line:84% And here, it tells you that this gene 00:38:40.440 --> 00:38:42.420 align:middle line:84% is in the coding sequence of the gene, 00:38:42.420 --> 00:38:48.270 align:middle line:84% and it actually causes an amino acid change at position 319. 00:38:48.270 --> 00:38:50.160 align:middle line:90% It affects protein domains-- 00:38:50.160 --> 00:38:51.990 align:middle line:84% this is the list of protein domains 00:38:51.990 --> 00:38:54.750 align:middle line:90% that are affected by the SNP. 00:38:54.750 --> 00:38:56.850 align:middle line:84% This is the list of some [? matters ?] 00:38:56.850 --> 00:39:01.800 align:middle line:84% of the investigators who have observed this SNP-- 00:39:01.800 --> 00:39:04.950 align:middle line:84% and it's a long list, so it means that this is definitely 00:39:04.950 --> 00:39:06.060 align:middle line:90% a true SNP. 00:39:06.060 --> 00:39:08.850 align:middle line:84% And finally under here, I wanted to show you, 00:39:08.850 --> 00:39:11.470 align:middle line:84% this data comes from TSC, and it's a frequency information 00:39:11.470 --> 00:39:11.970 align:middle line:90% data. 00:39:11.970 --> 00:39:16.650 align:middle line:84% So they sampled 41 individuals from a population 00:39:16.650 --> 00:39:19.620 align:middle line:84% of African-Americans, and they found that these are 00:39:19.620 --> 00:39:23.850 align:middle line:84% the frequencies for two alleles, 72% A, 20% G. 00:39:23.850 --> 00:39:25.930 align:middle line:84% And then they looked at a different population-- 00:39:25.930 --> 00:39:28.350 align:middle line:90% these are Caucasian, I think-- 00:39:28.350 --> 00:39:31.660 align:middle line:84% and they found very different alleles frequencies. 00:39:31.660 --> 00:39:35.010 align:middle line:84% So different that what was the minor allele in the first case 00:39:35.010 --> 00:39:36.870 align:middle line:90% is now the major allele. 00:39:36.870 --> 00:39:39.930 align:middle line:84% So this is a very clear demonstration 00:39:39.930 --> 00:39:42.540 align:middle line:84% of why it's important to know what population we're 00:39:42.540 --> 00:39:46.410 align:middle line:84% talking about when we study the frequency of a SNP. 00:39:46.410 --> 00:39:48.270 align:middle line:90% Because if you started-- 00:39:48.270 --> 00:39:52.200 align:middle line:84% if you believe these numbers, and then you 00:39:52.200 --> 00:39:54.960 align:middle line:84% try to run association studies in the SNP 00:39:54.960 --> 00:39:57.630 align:middle line:84% on a different population, you're 00:39:57.630 --> 00:39:59.550 align:middle line:84% going to find totally different numbers. 00:39:59.550 --> 00:40:01.680 align:middle line:84% And this doesn't have anything to do with disease-- 00:40:01.680 --> 00:40:05.670 align:middle line:84% you're going to just get results that are misleading, 00:40:05.670 --> 00:40:08.790 align:middle line:84% because you are not looking at the same population. 00:40:08.790 --> 00:40:11.220 align:middle line:84% And the baseline frequency of this SNP and the two 00:40:11.220 --> 00:40:14.440 align:middle line:90% populations is very different. 00:40:14.440 --> 00:40:21.495 align:middle line:84% So this is just to show that the advantages of having 00:40:21.495 --> 00:40:23.620 align:middle line:84% an integrated view that brings together information 00:40:23.620 --> 00:40:25.037 align:middle line:84% from different sources, and allows 00:40:25.037 --> 00:40:29.180 align:middle line:84% you to get a clear picture of what the SNP does 00:40:29.180 --> 00:40:32.410 align:middle line:84% and everything that is known about it. 00:40:32.410 --> 00:40:33.850 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:40:33.850 --> 00:40:36.040 align:middle line:90% ALBERTO RIVA: Excuse me? 00:40:36.040 --> 00:40:37.350 align:middle line:90% This one? 00:40:37.350 --> 00:40:46.000 align:middle line:84% Oh, well, it's just telling you that proteins have-- 00:40:46.000 --> 00:40:49.020 align:middle line:90% the sequence of a protein is-- 00:40:49.020 --> 00:40:52.960 align:middle line:84% well, it contains portions that are active domains, 00:40:52.960 --> 00:40:55.990 align:middle line:84% they are the portions of the protein that then physically do 00:40:55.990 --> 00:40:56.560 align:middle line:90% something. 00:40:56.560 --> 00:41:02.830 align:middle line:84% For example, this domain here is the extracellular domain, 00:41:02.830 --> 00:41:06.560 align:middle line:84% is the main that goes outside the cell. 00:41:06.560 --> 00:41:08.780 align:middle line:84% This is a calcium binding domain. 00:41:08.780 --> 00:41:11.110 align:middle line:84% So these are structures of the protein sequences 00:41:11.110 --> 00:41:16.475 align:middle line:84% that are known to have some function, 00:41:16.475 --> 00:41:18.350 align:middle line:84% they are important because they do something. 00:41:18.350 --> 00:41:21.880 align:middle line:84% And if you have a SNP that affects one of them, 00:41:21.880 --> 00:41:25.285 align:middle line:84% that SNP, in turn, might cause a protein to work-- 00:41:25.285 --> 00:41:30.160 align:middle line:90% 00:41:30.160 --> 00:41:32.470 align:middle line:84% it can change the function of a SNP-- 00:41:32.470 --> 00:41:33.477 align:middle line:90% of a protein. 00:41:33.477 --> 00:41:35.578 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:41:35.578 --> 00:41:37.120 align:middle line:84% ALBERTO RIVA: Well, this is not meant 00:41:37.120 --> 00:41:39.680 align:middle line:84% to be an accurate prediction of what the SNP does. 00:41:39.680 --> 00:41:44.260 align:middle line:84% And we get so many because all these domains are overlapping. 00:41:44.260 --> 00:41:47.230 align:middle line:84% And this information comes from Swiss [INAUDIBLE] database 00:41:47.230 --> 00:41:49.790 align:middle line:90% of protein information. 00:41:49.790 --> 00:41:59.970 align:middle line:84% And so you see for example, this first domain covers almost all 00:41:59.970 --> 00:42:01.288 align:middle line:90% of the protein. 00:42:01.288 --> 00:42:03.670 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:42:03.670 --> 00:42:05.130 align:middle line:90% 00:42:05.130 --> 00:42:06.270 align:middle line:90% ALBERTO RIVA: Excuse me? 00:42:06.270 --> 00:42:08.145 align:middle line:84% AUDIENCE: So six would be the maximum number? 00:42:08.145 --> 00:42:11.040 align:middle line:84% ALBERTO RIVA: No, no, it's just that these domains can 00:42:11.040 --> 00:42:13.440 align:middle line:84% be overlapping, just because the Swiss people, 00:42:13.440 --> 00:42:15.730 align:middle line:84% they annotate the protein sequence saying OK, 00:42:15.730 --> 00:42:18.480 align:middle line:84% from here to here, we know that this happens. 00:42:18.480 --> 00:42:24.460 align:middle line:90% But-- well, sometimes they-- 00:42:24.460 --> 00:42:26.520 align:middle line:84% well, there are some domains that 00:42:26.520 --> 00:42:28.830 align:middle line:84% cover the entire protein, or half of the protein, 00:42:28.830 --> 00:42:32.250 align:middle line:84% just because, for example, in this case, the extracellular 00:42:32.250 --> 00:42:36.090 align:middle line:84% domain, it means that this portion of the protein 00:42:36.090 --> 00:42:38.730 align:middle line:90% is extracellular. 00:42:38.730 --> 00:42:41.010 align:middle line:84% And then inside that domain, you can 00:42:41.010 --> 00:42:42.870 align:middle line:84% have other subdomains like all these 00:42:42.870 --> 00:42:45.260 align:middle line:90% that have other characteristics. 00:42:45.260 --> 00:42:47.970 align:middle line:84% So I'm just reporting here a list of all the domains 00:42:47.970 --> 00:42:52.908 align:middle line:84% that contain that location, but they can be overlapping. 00:42:52.908 --> 00:42:55.200 align:middle line:84% So it doesn't necessarily mean that the SNP affects all 00:42:55.200 --> 00:42:56.760 align:middle line:90% of them in some meaningful way. 00:42:56.760 --> 00:42:59.500 align:middle line:90% 00:42:59.500 --> 00:43:01.900 align:middle line:84% This one is probably the only one 00:43:01.900 --> 00:43:06.070 align:middle line:84% that could be affected by the presence of a SNP. 00:43:06.070 --> 00:43:08.020 align:middle line:84% Because it's a binding domain, so it 00:43:08.020 --> 00:43:11.960 align:middle line:84% might be that it doesn't work anymore as a binding domain. 00:43:11.960 --> 00:43:14.080 align:middle line:84% So don't get confused by this place. 00:43:14.080 --> 00:43:16.880 align:middle line:84% Just a list of Swiss [INAUDIBLE] domains 00:43:16.880 --> 00:43:18.130 align:middle line:90% that include that location. 00:43:18.130 --> 00:43:23.510 align:middle line:90% 00:43:23.510 --> 00:43:29.770 align:middle line:84% Now, the next step I'm going to talk about genes again. 00:43:29.770 --> 00:43:32.890 align:middle line:84% And the starting point, when we talk about genes, 00:43:32.890 --> 00:43:34.270 align:middle line:90% it's LocusLink. 00:43:34.270 --> 00:43:42.033 align:middle line:84% LocusLink is a curated directory of genes from 13 organisms. 00:43:42.033 --> 00:43:43.700 align:middle line:84% The word curated here is very important. 00:43:43.700 --> 00:43:45.700 align:middle line:84% So genes are discovered either experimentally, 00:43:45.700 --> 00:43:50.680 align:middle line:84% or by programs like GenScan that look at the DNA sequence 00:43:50.680 --> 00:43:54.670 align:middle line:84% and tell you where a gene might be. 00:43:54.670 --> 00:43:58.300 align:middle line:84% Then the gene has to be studied in order 00:43:58.300 --> 00:44:03.250 align:middle line:84% to know what it does, what all its relationship to other genes 00:44:03.250 --> 00:44:05.200 align:middle line:90% and biological processes are. 00:44:05.200 --> 00:44:08.560 align:middle line:84% So LocusLink is basically a repository 00:44:08.560 --> 00:44:10.810 align:middle line:84% of information about genes, and it collects everything 00:44:10.810 --> 00:44:13.420 align:middle line:90% that is known about the genes. 00:44:13.420 --> 00:44:17.922 align:middle line:84% So they say their central function 00:44:17.922 --> 00:44:20.380 align:middle line:84% is to establish an accurate connection between the defining 00:44:20.380 --> 00:44:22.243 align:middle line:84% sequence for locus and other descriptors. 00:44:22.243 --> 00:44:24.160 align:middle line:84% It basically means, you have a stretch of DNA, 00:44:24.160 --> 00:44:27.130 align:middle line:84% you know that there is a gene, let's collect everything 00:44:27.130 --> 00:44:30.380 align:middle line:90% that is known about that gene. 00:44:30.380 --> 00:44:32.710 align:middle line:84% So it gives you information about the sequence, 00:44:32.710 --> 00:44:35.470 align:middle line:84% itself, about the functions of the gene, 00:44:35.470 --> 00:44:39.190 align:middle line:84% links to other databases, about the gene, different names 00:44:39.190 --> 00:44:41.710 align:middle line:84% for the gene, phenotypes that are known to be associated 00:44:41.710 --> 00:44:44.380 align:middle line:84% with that gene, homologous to other genes 00:44:44.380 --> 00:44:47.290 align:middle line:84% in the same organism, or in different organisms, 00:44:47.290 --> 00:44:49.990 align:middle line:84% the location of this gene in several different maps. 00:44:49.990 --> 00:44:53.810 align:middle line:84% This is all information that you can find in LocusLink. 00:44:53.810 --> 00:44:55.060 align:middle line:90% And the most important thing-- 00:44:55.060 --> 00:44:57.890 align:middle line:90% 00:44:57.890 --> 00:45:00.130 align:middle line:84% at least from our point of view-- 00:45:00.130 --> 00:45:04.150 align:middle line:84% is that LocusLink provides a nomenclature of genes. 00:45:04.150 --> 00:45:06.710 align:middle line:84% No LocusLink assigns a name to each gene, 00:45:06.710 --> 00:45:10.000 align:middle line:84% and if you stick to that name, then you're 00:45:10.000 --> 00:45:12.790 align:middle line:84% sure that everybody knows what you're talking about. 00:45:12.790 --> 00:45:15.580 align:middle line:84% Because this, again, it might seem a trivial problem, 00:45:15.580 --> 00:45:20.770 align:middle line:84% but for historical reasons, in many, many cases, 00:45:20.770 --> 00:45:22.300 align:middle line:84% genes have lots of different names, 00:45:22.300 --> 00:45:25.387 align:middle line:84% even if it's the same gene, people 00:45:25.387 --> 00:45:27.220 align:middle line:84% have been calling them with different names, 00:45:27.220 --> 00:45:29.988 align:middle line:84% and it's a mess when you try to out which gene is which. 00:45:29.988 --> 00:45:31.780 align:middle line:84% If you stick to the LocusLink nomenclature, 00:45:31.780 --> 00:45:37.270 align:middle line:84% then at least you have one way of naming genes, and that's it. 00:45:37.270 --> 00:45:39.230 align:middle line:84% So it gives a name, it gives a number, 00:45:39.230 --> 00:45:41.290 align:middle line:84% and you can use these as identifiers 00:45:41.290 --> 00:45:43.270 align:middle line:84% to look up your gene in other databases, 00:45:43.270 --> 00:45:45.670 align:middle line:84% if they use the same nomenclature. 00:45:45.670 --> 00:45:48.700 align:middle line:84% And of course, again, it's part of the NCBI cluster, 00:45:48.700 --> 00:45:52.213 align:middle line:84% and all NCBI resources use this way of naming genes. 00:45:52.213 --> 00:45:54.130 align:middle line:84% Then, unfortunately, there are other resources 00:45:54.130 --> 00:45:57.280 align:middle line:84% that we'll mention later that use 00:45:57.280 --> 00:46:01.450 align:middle line:84% a different way of naming genes, and this 00:46:01.450 --> 00:46:03.130 align:middle line:84% makes things very difficult when you're 00:46:03.130 --> 00:46:06.610 align:middle line:84% trying to build programs to integrate information 00:46:06.610 --> 00:46:09.880 align:middle line:84% from different places, because it's very, very 00:46:09.880 --> 00:46:13.420 align:middle line:84% hard to know exactly how to reconcile 00:46:13.420 --> 00:46:15.850 align:middle line:90% different ways of naming genes. 00:46:15.850 --> 00:46:19.210 align:middle line:84% Again, it might seem a trivial problem, but it's not. 00:46:19.210 --> 00:46:23.130 align:middle line:84% And it's also complicated by the fact 00:46:23.130 --> 00:46:29.570 align:middle line:84% that, as I was saying before, genes 00:46:29.570 --> 00:46:32.010 align:middle line:84% may appear in several different forms-- 00:46:32.010 --> 00:46:36.110 align:middle line:84% there are variants of the same gene, 00:46:36.110 --> 00:46:38.430 align:middle line:84% there are genes that are very similar to each other. 00:46:38.430 --> 00:46:40.250 align:middle line:84% So sometimes they are considered to be the same genes, 00:46:40.250 --> 00:46:41.167 align:middle line:90% sometimes they're not. 00:46:41.167 --> 00:46:46.640 align:middle line:84% And all these are things that make the naming genes 00:46:46.640 --> 00:46:52.385 align:middle line:84% kind of a complex and not a deterministic task. 00:46:52.385 --> 00:46:55.160 align:middle line:90% 00:46:55.160 --> 00:46:57.770 align:middle line:84% So UniGene is another resource at NCBI 00:46:57.770 --> 00:46:59.545 align:middle line:84% that takes a slightly different approach. 00:46:59.545 --> 00:47:03.380 align:middle line:90% 00:47:03.380 --> 00:47:08.210 align:middle line:84% It's an attempt at collecting all the GenBank sequences that 00:47:08.210 --> 00:47:12.950 align:middle line:84% refer to a region of the genome where a gene is known to be. 00:47:12.950 --> 00:47:18.110 align:middle line:84% So essentially, if we know that a certain region 00:47:18.110 --> 00:47:19.970 align:middle line:84% of our chromosome contains a gene, 00:47:19.970 --> 00:47:23.120 align:middle line:84% then we can go into GenBank and look at all the sequences 00:47:23.120 --> 00:47:24.300 align:middle line:90% that fall into that region. 00:47:24.300 --> 00:47:30.800 align:middle line:84% So all the sequences ultimately come from that gene or part 00:47:30.800 --> 00:47:32.030 align:middle line:90% of that gene. 00:47:32.030 --> 00:47:34.700 align:middle line:84% And UniGene puts them all together in one cluster. 00:47:34.700 --> 00:47:42.110 align:middle line:84% And then tries to provide a description of why 00:47:42.110 --> 00:47:44.510 align:middle line:84% all these sequences-- of description 00:47:44.510 --> 00:47:47.920 align:middle line:84% of the features of all these sequences. 00:47:47.920 --> 00:47:50.420 align:middle line:84% So they're all similar, they all come from the same location 00:47:50.420 --> 00:47:53.510 align:middle line:84% from the same region of the genome. 00:47:53.510 --> 00:47:55.647 align:middle line:84% But they might represent multiple forms 00:47:55.647 --> 00:47:57.230 align:middle line:84% of the same genes, so they're probably 00:47:57.230 --> 00:47:58.825 align:middle line:90% not identical to each other. 00:47:58.825 --> 00:48:00.450 align:middle line:84% They might come from different tissues, 00:48:00.450 --> 00:48:05.880 align:middle line:84% so they might have different properties, and so on. 00:48:05.880 --> 00:48:09.230 align:middle line:84% And again, this is the URL for UniGene. 00:48:09.230 --> 00:48:15.470 align:middle line:84% It includes information for 38 organisms. 00:48:15.470 --> 00:48:17.220 align:middle line:84% And I think that one year ago, this number 00:48:17.220 --> 00:48:18.420 align:middle line:90% was something like 14. 00:48:18.420 --> 00:48:21.240 align:middle line:90% So it's growing very fast. 00:48:21.240 --> 00:48:25.240 align:middle line:84% And the interesting thing is, this is an automated process. 00:48:25.240 --> 00:48:27.360 align:middle line:84% So LocusLink is a curated directory, 00:48:27.360 --> 00:48:32.430 align:middle line:84% means that there are people who spend their days going 00:48:32.430 --> 00:48:36.330 align:middle line:84% through gene records and adding information, checking 00:48:36.330 --> 00:48:37.640 align:middle line:90% it, correcting it. 00:48:37.640 --> 00:48:41.340 align:middle line:84% UniGene is an automated system, so it's actually 00:48:41.340 --> 00:48:47.460 align:middle line:84% an automated procedure that looks at all the GenBank 00:48:47.460 --> 00:48:51.030 align:middle line:84% sequences and tries to build these clusters based 00:48:51.030 --> 00:48:58.340 align:middle line:84% on the location of the sequences. 00:48:58.340 --> 00:49:01.310 align:middle line:90% 00:49:01.310 --> 00:49:03.050 align:middle line:84% I've mentioned the fact that it's 00:49:03.050 --> 00:49:06.650 align:middle line:84% interesting to study homologies between genes 00:49:06.650 --> 00:49:07.650 align:middle line:90% and different organisms. 00:49:07.650 --> 00:49:13.685 align:middle line:84% So HomoloGene is a database of all orthologs. 00:49:13.685 --> 00:49:18.740 align:middle line:84% So what it does, they take all the sequences in GenBank, 00:49:18.740 --> 00:49:22.100 align:middle line:84% they compare each sequence with all the other sequences 00:49:22.100 --> 00:49:28.450 align:middle line:84% in GenBank, at least in a set of organisms. 00:49:28.450 --> 00:49:31.480 align:middle line:84% And if they find a good match between the two sequences, 00:49:31.480 --> 00:49:36.730 align:middle line:84% then this pair is added to the HomoloGene database. 00:49:36.730 --> 00:49:41.830 align:middle line:84% So right now it encompasses 25 organisms. 00:49:41.830 --> 00:49:47.170 align:middle line:84% And in these 25 organisms, they have 470,000 ortholog pairs-- 00:49:47.170 --> 00:49:49.480 align:middle line:84% so pairs of genes from different organisms that are 00:49:49.480 --> 00:49:52.270 align:middle line:90% highly similar to each other. 00:49:52.270 --> 00:49:55.640 align:middle line:84% All these are put into the database. 00:49:55.640 --> 00:49:58.120 align:middle line:84% And then if you find there are three organisms that 00:49:58.120 --> 00:50:03.730 align:middle line:84% share a similarity relationship, then this, in turn, 00:50:03.730 --> 00:50:09.340 align:middle line:84% is marked, because it means that you're finding a match that 00:50:09.340 --> 00:50:10.550 align:middle line:90% has an even higher quality. 00:50:10.550 --> 00:50:14.080 align:middle line:84% So if you find that organism A shares a gene with organism B, 00:50:14.080 --> 00:50:18.340 align:middle line:84% and B shares it with C, If then you find a C shares it with A, 00:50:18.340 --> 00:50:20.290 align:middle line:84% then you've built what they call a triplet. 00:50:20.290 --> 00:50:26.020 align:middle line:84% And that's a confirmation that actually this gene 00:50:26.020 --> 00:50:29.080 align:middle line:84% might be really the same gene that is conserved 00:50:29.080 --> 00:50:32.990 align:middle line:90% across all these organisms. 00:50:32.990 --> 00:50:35.320 align:middle line:84% This one is partly curated, partly calculated. 00:50:35.320 --> 00:50:38.200 align:middle line:84% So they have an automated procedure 00:50:38.200 --> 00:50:41.620 align:middle line:84% that looks at sequence similarity using all 00:50:41.620 --> 00:50:43.762 align:middle line:90% the many algorithms to do that. 00:50:43.762 --> 00:50:45.220 align:middle line:84% And they give the similarity score. 00:50:45.220 --> 00:50:48.205 align:middle line:90% And then they have a subset-- 00:50:48.205 --> 00:50:49.330 align:middle line:90% this is not mentioned here. 00:50:49.330 --> 00:50:53.440 align:middle line:84% But then, most of these entries in [INAUDIBLE] are also 00:50:53.440 --> 00:50:57.070 align:middle line:84% manually curated to make sure that they are really-- 00:50:57.070 --> 00:51:00.130 align:middle line:90% 00:51:00.130 --> 00:51:01.805 align:middle line:84% that they're really similar genes. 00:51:01.805 --> 00:51:04.130 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:51:04.130 --> 00:51:06.350 align:middle line:90% 00:51:06.350 --> 00:51:08.240 align:middle line:84% ALBERTO RIVA: I think it's a Swift-- 00:51:08.240 --> 00:51:09.920 align:middle line:84% no, it's part of the [? blast ?] score. 00:51:09.920 --> 00:51:13.025 align:middle line:84% Yeah, they have a threshold of something-- 00:51:13.025 --> 00:51:13.790 align:middle line:90% I don't remember. 00:51:13.790 --> 00:51:16.940 align:middle line:84% But they give it the score in addition 00:51:16.940 --> 00:51:18.440 align:middle line:90% to all the other information. 00:51:18.440 --> 00:51:21.880 align:middle line:90% 00:51:21.880 --> 00:51:28.960 align:middle line:84% OK, Ensemble-- this is not part of NCBI. 00:51:28.960 --> 00:51:34.083 align:middle line:84% This is something that comes from Europe, from the EMBL-- 00:51:34.083 --> 00:51:35.500 align:middle line:84% European Bioinformatics Institute, 00:51:35.500 --> 00:51:37.450 align:middle line:90% the Sanger Institute. 00:51:37.450 --> 00:51:43.660 align:middle line:84% It's something that is pretty similar to LocusLink in scope. 00:51:43.660 --> 00:51:47.290 align:middle line:84% Again, it's a software system for the automated annotation 00:51:47.290 --> 00:51:49.120 align:middle line:84% of genomes-- it's basically means 00:51:49.120 --> 00:51:51.850 align:middle line:84% it's a system that discovers genes and tries 00:51:51.850 --> 00:51:55.210 align:middle line:84% to find as much information as possible about these genes. 00:51:55.210 --> 00:51:59.410 align:middle line:84% And then all the information is available through a search 00:51:59.410 --> 00:52:01.390 align:middle line:90% interface. 00:52:01.390 --> 00:52:04.390 align:middle line:84% It's limited to 10 organisms, but it 00:52:04.390 --> 00:52:08.500 align:middle line:84% provides a lot of information about the genes in this-- 00:52:08.500 --> 00:52:09.418 align:middle line:90% about this organism. 00:52:09.418 --> 00:52:11.710 align:middle line:84% So it provides information about genes, about proteins, 00:52:11.710 --> 00:52:15.400 align:middle line:84% diseases, SNPs, cross-species analysis, microarray data. 00:52:15.400 --> 00:52:16.810 align:middle line:84% So it's essentially a combination 00:52:16.810 --> 00:52:23.560 align:middle line:84% of LocusLink, dbSNP, HomoloGene, and a few other things. 00:52:23.560 --> 00:52:28.040 align:middle line:84% It has a very powerful data access interface. 00:52:28.040 --> 00:52:30.980 align:middle line:84% It's actually very, very nice, very easy to use. 00:52:30.980 --> 00:52:35.230 align:middle line:84% So you can do queries on this huge database 00:52:35.230 --> 00:52:39.110 align:middle line:90% in a relatively simple way. 00:52:39.110 --> 00:52:42.052 align:middle line:84% One of the biggest problems with this system, 00:52:42.052 --> 00:52:43.510 align:middle line:84% at least from our point of view, is 00:52:43.510 --> 00:52:47.248 align:middle line:84% that it uses its own way of naming genes. 00:52:47.248 --> 00:52:49.540 align:middle line:84% This is essentially what I was referring to before when 00:52:49.540 --> 00:52:52.390 align:middle line:84% I was saying that not everybody uses the LocusLink 00:52:52.390 --> 00:52:53.890 align:middle line:90% way of naming genes. 00:52:53.890 --> 00:52:59.050 align:middle line:84% They have their own alternative scheme for naming genes. 00:52:59.050 --> 00:53:03.360 align:middle line:84% And going from one to the other is sometimes tricky. 00:53:03.360 --> 00:53:05.110 align:middle line:84% There are links between the two databases. 00:53:05.110 --> 00:53:06.970 align:middle line:90% But of course it's not-- 00:53:06.970 --> 00:53:09.745 align:middle line:84% they don't necessarily match very well. 00:53:09.745 --> 00:53:13.650 align:middle line:90% 00:53:13.650 --> 00:53:14.320 align:middle line:90% What else? 00:53:14.320 --> 00:53:17.100 align:middle line:84% OK, and finally, a few words about gene regulation. 00:53:17.100 --> 00:53:22.630 align:middle line:90% 00:53:22.630 --> 00:53:30.160 align:middle line:84% So gene regulation-- of course, it's almost needless to say, 00:53:30.160 --> 00:53:32.440 align:middle line:84% it's an extremely complex mechanism. 00:53:32.440 --> 00:53:34.450 align:middle line:84% Our understanding of how gene regulation works 00:53:34.450 --> 00:53:36.400 align:middle line:90% is still very limited. 00:53:36.400 --> 00:53:42.190 align:middle line:84% When you hear about microarrays, about the concept of gene 00:53:42.190 --> 00:53:45.760 align:middle line:84% expression as measured by microarrays, 00:53:45.760 --> 00:53:51.580 align:middle line:84% gene expression is the most visible consequence 00:53:51.580 --> 00:53:56.440 align:middle line:84% of everything that is in this complex mechanism. 00:53:56.440 --> 00:54:01.150 align:middle line:84% So what you see is that under certain conditions, 00:54:01.150 --> 00:54:04.750 align:middle line:84% a certain set of genes is highly regulated, is highly expressed, 00:54:04.750 --> 00:54:08.248 align:middle line:90% or under expressed, and so on. 00:54:08.248 --> 00:54:09.790 align:middle line:84% But this is a consequence of the fact 00:54:09.790 --> 00:54:15.310 align:middle line:84% that there is a very complex machinery behind it that 00:54:15.310 --> 00:54:18.520 align:middle line:84% determines which genes are active or not, and how much, 00:54:18.520 --> 00:54:21.580 align:middle line:90% in different conditions. 00:54:21.580 --> 00:54:23.440 align:middle line:84% And this is actually a system that 00:54:23.440 --> 00:54:26.140 align:middle line:84% integrates a lot of different factors 00:54:26.140 --> 00:54:28.660 align:middle line:84% that might include the following, 00:54:28.660 --> 00:54:30.190 align:middle line:90% in no particular order-- 00:54:30.190 --> 00:54:35.110 align:middle line:84% The tissue, we know very well that the set of genes 00:54:35.110 --> 00:54:36.700 align:middle line:84% that are expressed in one tissue is 00:54:36.700 --> 00:54:38.992 align:middle line:84% very different from the set of genes that are expressed 00:54:38.992 --> 00:54:40.330 align:middle line:90% in another tissue. 00:54:40.330 --> 00:54:42.460 align:middle line:84% Developmental stage, genes that are expressed 00:54:42.460 --> 00:54:45.610 align:middle line:84% during the development of the embryo, for example, 00:54:45.610 --> 00:54:51.280 align:middle line:84% are not the same that are expressed in an adult organism. 00:54:51.280 --> 00:54:53.980 align:middle line:84% The time-- the time can mean either a time of day, 00:54:53.980 --> 00:54:55.990 align:middle line:84% for the case like the circadian rhythm, 00:54:55.990 --> 00:54:58.032 align:middle line:84% there are genes that are expressed in the morning 00:54:58.032 --> 00:54:59.920 align:middle line:90% and not in the evening. 00:54:59.920 --> 00:55:03.820 align:middle line:84% Or time at a larger scale, there are 00:55:03.820 --> 00:55:06.040 align:middle line:84% processes that take years to complete, 00:55:06.040 --> 00:55:08.905 align:middle line:90% like puberty, for example. 00:55:08.905 --> 00:55:12.870 align:middle line:90% 00:55:12.870 --> 00:55:16.200 align:middle line:84% So there is this regulation mechanism 00:55:16.200 --> 00:55:24.240 align:middle line:84% is able to work at very different temporal resolutions. 00:55:24.240 --> 00:55:29.110 align:middle line:84% External signals, of course, all response to external stimuli. 00:55:29.110 --> 00:55:31.080 align:middle line:84% And it also depends on the expression state 00:55:31.080 --> 00:55:33.450 align:middle line:84% of any number of other genes, because genes 00:55:33.450 --> 00:55:37.060 align:middle line:84% regulate each other through feedback loops and so on. 00:55:37.060 --> 00:55:41.720 align:middle line:84% So again, it's a very complex system. 00:55:41.720 --> 00:55:46.560 align:middle line:84% We're slowly working to try to understand how it works. 00:55:46.560 --> 00:55:50.775 align:middle line:84% So what we have for now is some understanding 00:55:50.775 --> 00:55:53.150 align:middle line:84% of what transcription factors-- the transcription factors 00:55:53.150 --> 00:55:57.140 align:middle line:84% are proteins that bind to the upstream regions of the genes, 00:55:57.140 --> 00:56:00.410 align:middle line:84% and are able to control their expression, their activity. 00:56:00.410 --> 00:56:03.740 align:middle line:84% Because transcription factor, because usually the most common 00:56:03.740 --> 00:56:07.290 align:middle line:84% case-- these bind to the promoter region of a gene, 00:56:07.290 --> 00:56:09.710 align:middle line:84% they combine with each other, forming complexes. 00:56:09.710 --> 00:56:13.160 align:middle line:84% And these complexes then activate the transcription 00:56:13.160 --> 00:56:19.800 align:middle line:84% machinery, that then gives rise to what starts everything else. 00:56:19.800 --> 00:56:21.590 align:middle line:84% And in the end, you get the gene is 00:56:21.590 --> 00:56:24.380 align:middle line:84% expressed, because it was latent in the protein is produced. 00:56:24.380 --> 00:56:27.500 align:middle line:90% 00:56:27.500 --> 00:56:29.750 align:middle line:84% And the transcription factors, as I was saying, 00:56:29.750 --> 00:56:31.108 align:middle line:90% don't act alone. 00:56:31.108 --> 00:56:32.900 align:middle line:84% They have to interact with the target gene, 00:56:32.900 --> 00:56:35.150 align:middle line:84% but they also interact with each other 00:56:35.150 --> 00:56:36.380 align:middle line:90% in a combinatorial fashion. 00:56:36.380 --> 00:56:39.650 align:middle line:84% What this means is that, looking at the individual transcription 00:56:39.650 --> 00:56:42.543 align:middle line:84% factor is usually not sufficient to understand 00:56:42.543 --> 00:56:43.460 align:middle line:90% what it's going to do. 00:56:43.460 --> 00:56:45.757 align:middle line:84% Because the same transcription factor 00:56:45.757 --> 00:56:47.840 align:middle line:84% in different combinations with other transcription 00:56:47.840 --> 00:56:51.720 align:middle line:84% factors might have different roles. 00:56:51.720 --> 00:56:54.380 align:middle line:84% So what we need to look at is the pattern 00:56:54.380 --> 00:56:57.560 align:middle line:84% of transcription factors that binds to a certain gene. 00:56:57.560 --> 00:56:59.360 align:middle line:84% And that, in turn, will determine 00:56:59.360 --> 00:57:07.050 align:middle line:84% the spatial, temporal, dependent expression of the target gene. 00:57:07.050 --> 00:57:11.877 align:middle line:84% And again, we are still doing the very early steps 00:57:11.877 --> 00:57:13.460 align:middle line:84% in the process of trying to understand 00:57:13.460 --> 00:57:18.930 align:middle line:84% how these patterns are actually structured, how they work. 00:57:18.930 --> 00:57:22.060 align:middle line:90% So what we-- 00:57:22.060 --> 00:57:24.760 align:middle line:90% OK, sorry. 00:57:24.760 --> 00:57:27.380 align:middle line:84% So the first step, again, we're moving the first steps. 00:57:27.380 --> 00:57:28.810 align:middle line:84% The first thing you need to do is 00:57:28.810 --> 00:57:31.960 align:middle line:84% need to be able to reliably identify 00:57:31.960 --> 00:57:38.230 align:middle line:84% which transcription factors bind to a given gene, and where, 00:57:38.230 --> 00:57:42.430 align:middle line:84% exactly, in the promoter region of the gene they bind. 00:57:42.430 --> 00:57:45.550 align:middle line:84% And transcription factors bind to locations 00:57:45.550 --> 00:57:48.430 align:middle line:84% that are called transcription factor binding sites. 00:57:48.430 --> 00:57:50.080 align:middle line:84% They're small stretches of DNA that 00:57:50.080 --> 00:57:51.800 align:middle line:90% are recognized by the factor. 00:57:51.800 --> 00:57:55.720 align:middle line:84% And so if you know where the binding sites are, 00:57:55.720 --> 00:58:01.990 align:middle line:84% you have a first idea of what factors bind to this gene, 00:58:01.990 --> 00:58:04.390 align:middle line:84% and how they may be arranged spatially. 00:58:04.390 --> 00:58:06.400 align:middle line:84% So if you know that two factors have 00:58:06.400 --> 00:58:09.820 align:middle line:84% to interact with each other, probably their binding sites 00:58:09.820 --> 00:58:11.860 align:middle line:84% will have to be close to each other. 00:58:11.860 --> 00:58:14.530 align:middle line:84% Or, at least, let's say if you find two binding sites that 00:58:14.530 --> 00:58:16.697 align:middle line:84% are close to each other, there is a very high chance 00:58:16.697 --> 00:58:18.280 align:middle line:90% that two factors will interact. 00:58:18.280 --> 00:58:20.860 align:middle line:84% And it might be that when they interact, 00:58:20.860 --> 00:58:22.743 align:middle line:90% they act in a certain way. 00:58:22.743 --> 00:58:24.910 align:middle line:84% Where they don't interact because they're far apart, 00:58:24.910 --> 00:58:26.270 align:middle line:90% and they act in a different way. 00:58:26.270 --> 00:58:28.955 align:middle line:84% And so knowing the map of binding sites, 00:58:28.955 --> 00:58:30.580 align:middle line:84% and the promoter of a gene is something 00:58:30.580 --> 00:58:36.790 align:middle line:84% that can give you some initial information 00:58:36.790 --> 00:58:39.560 align:middle line:90% that it can build on. 00:58:39.560 --> 00:58:44.650 align:middle line:84% It's still something that's very hard to do computationally. 00:58:44.650 --> 00:58:50.110 align:middle line:84% The ways that [? truth ?] that people 00:58:50.110 --> 00:58:52.030 align:middle line:84% have been using to do this are usually 00:58:52.030 --> 00:58:53.210 align:middle line:90% based on pattern matching. 00:58:53.210 --> 00:58:57.040 align:middle line:84% So the binding site, as I said, is a small stretch of DNA, 00:58:57.040 --> 00:59:00.650 align:middle line:84% usually goes from five to about 20 or 25 base pairs. 00:59:00.650 --> 00:59:02.410 align:middle line:90% So they're really short. 00:59:02.410 --> 00:59:06.880 align:middle line:84% And they're characterized by [? concentric ?] sequences-- 00:59:06.880 --> 00:59:10.760 align:middle line:84% in general, they are not very [? concern-- ?] 00:59:10.760 --> 00:59:11.750 align:middle line:90% not very precise. 00:59:11.750 --> 00:59:16.330 align:middle line:84% So it's essentially impossible to look at a piece of DNA 00:59:16.330 --> 00:59:19.780 align:middle line:84% and say, well, OK, I'm sure that this location here 00:59:19.780 --> 00:59:22.190 align:middle line:90% is a binding site. 00:59:22.190 --> 00:59:24.910 align:middle line:84% So you can try using deterministic methods, 00:59:24.910 --> 00:59:31.480 align:middle line:84% just looking for instances of the motifs. 00:59:31.480 --> 00:59:35.010 align:middle line:84% Sorry, thought I had something on this. 00:59:35.010 --> 00:59:41.565 align:middle line:84% But you can look for instances of the motifs using 00:59:41.565 --> 00:59:44.190 align:middle line:84% either deterministic methods, or probabilistic methods, pattern 00:59:44.190 --> 00:59:48.330 align:middle line:84% matching, there is lots of things that you can try. 00:59:48.330 --> 00:59:57.750 align:middle line:84% And in almost all cases, people rely on TransFIC-- 00:59:57.750 --> 01:00:00.180 align:middle line:84% TransFIC is the largest available database 01:00:00.180 --> 01:00:02.035 align:middle line:90% about transcription factors. 01:00:02.035 --> 01:00:04.410 align:middle line:84% It's a database that provides information on the factors, 01:00:04.410 --> 01:00:05.400 align:middle line:90% themselves. 01:00:05.400 --> 01:00:08.790 align:middle line:84% It provides examples of their binding sites. 01:00:08.790 --> 01:00:12.780 align:middle line:84% And it provides descriptions of their interactions with genes. 01:00:12.780 --> 01:00:15.210 align:middle line:84% And the important thing is that most 01:00:15.210 --> 01:00:18.690 align:middle line:84% of the information in TransFIC is experimentally validated. 01:00:18.690 --> 01:00:20.495 align:middle line:84% So for example, the binding sites, 01:00:20.495 --> 01:00:21.870 align:middle line:84% these are binding sites that have 01:00:21.870 --> 01:00:23.100 align:middle line:90% been observed experimentally. 01:00:23.100 --> 01:00:28.530 align:middle line:84% So you can actually trust the fact 01:00:28.530 --> 01:00:30.870 align:middle line:84% that particular piece of sequence they 01:00:30.870 --> 01:00:33.240 align:middle line:84% give you is the binding site for the transcription 01:00:33.240 --> 01:00:34.170 align:middle line:90% factor in question. 01:00:34.170 --> 01:00:37.490 align:middle line:90% 01:00:37.490 --> 01:00:45.250 align:middle line:84% And so in the end, without going into too much detail, 01:00:45.250 --> 01:00:47.590 align:middle line:84% what you can do is can take these binding sites, 01:00:47.590 --> 01:00:50.080 align:middle line:84% you can use them to train your favorite pattern matching 01:00:50.080 --> 01:00:55.360 align:middle line:84% method, and then you can try scanning new sequences looking 01:00:55.360 --> 01:00:56.200 align:middle line:90% for binding sites. 01:00:56.200 --> 01:00:59.230 align:middle line:84% And this is one of the things that we're currently 01:00:59.230 --> 01:01:02.350 align:middle line:84% working on at CHB-- there are various ways of doing this. 01:01:02.350 --> 01:01:06.742 align:middle line:84% And again, it's a rather difficult problem, 01:01:06.742 --> 01:01:08.200 align:middle line:84% from a computational point of view, 01:01:08.200 --> 01:01:11.140 align:middle line:84% because these patterns that you have to look for 01:01:11.140 --> 01:01:13.610 align:middle line:90% are not very specific. 01:01:13.610 --> 01:01:18.280 align:middle line:90% They are not very clear. 01:01:18.280 --> 01:01:21.970 align:middle line:84% On the other hand, doing it computationally-- sorry, 01:01:21.970 --> 01:01:26.180 align:middle line:84% doing it experimentally is very slow, very expensive. 01:01:26.180 --> 01:01:28.380 align:middle line:84% So you can only do that for a small number of genes, 01:01:28.380 --> 01:01:29.800 align:middle line:90% and a small number of factors. 01:01:29.800 --> 01:01:32.048 align:middle line:84% If you have a method like this, if you 01:01:32.048 --> 01:01:34.340 align:middle line:84% have a computational meter to detect binding sites that 01:01:34.340 --> 01:01:36.340 align:middle line:84% works well, then you can think about doing 01:01:36.340 --> 01:01:38.710 align:middle line:84% this on a large scale, looking for example, 01:01:38.710 --> 01:01:41.230 align:middle line:84% all the binding sites for a certain factor in all 01:01:41.230 --> 01:01:42.460 align:middle line:90% the human genes. 01:01:42.460 --> 01:01:46.090 align:middle line:84% And that will give you a very interesting picture 01:01:46.090 --> 01:01:48.745 align:middle line:84% of everything that might be regulated by that factor. 01:01:48.745 --> 01:01:52.310 align:middle line:90% 01:01:52.310 --> 01:01:53.338 align:middle line:90% So we're not there yet. 01:01:53.338 --> 01:01:55.630 align:middle line:84% This is one of the things we are working on in our lab. 01:01:55.630 --> 01:02:03.940 align:middle line:84% And it's going to take a lot of work 01:02:03.940 --> 01:02:06.040 align:middle line:84% but the rewards are potentially very interesting 01:02:06.040 --> 01:02:08.740 align:middle line:84% because this is something that will then allow you, 01:02:08.740 --> 01:02:15.130 align:middle line:84% if it works, to automatically build 01:02:15.130 --> 01:02:18.680 align:middle line:84% the networks that describe how genes regulate each other. 01:02:18.680 --> 01:02:22.840 align:middle line:84% And that is something that, of course, has 01:02:22.840 --> 01:02:24.460 align:middle line:90% a lot of potential interest. 01:02:24.460 --> 01:02:28.375 align:middle line:90% 01:02:28.375 --> 01:02:30.000 align:middle line:84% OK, we've talked about gene expression, 01:02:30.000 --> 01:02:33.060 align:middle line:90% we talked about microarrays. 01:02:33.060 --> 01:02:37.440 align:middle line:84% You might have already heard about these things, 01:02:37.440 --> 01:02:42.570 align:middle line:84% but I was just going to list the main sources 01:02:42.570 --> 01:02:45.570 align:middle line:84% of available microarray data, public microarray data. 01:02:45.570 --> 01:02:50.580 align:middle line:84% So for example, again at NCBI, GEO 01:02:50.580 --> 01:02:52.290 align:middle line:84% is a database, Gene Expression Omnibus, 01:02:52.290 --> 01:02:56.040 align:middle line:84% is a database of gene expression and hybridization array data. 01:02:56.040 --> 01:03:00.030 align:middle line:84% It offers 12,000 experiments, essentially, 01:03:00.030 --> 01:03:02.850 align:middle line:84% 12,000 hybridization experiments on over 500 . 01:03:02.850 --> 01:03:06.060 align:middle line:84% Platforms so if you're interested in doing 01:03:06.060 --> 01:03:08.730 align:middle line:84% some form of data analysis on microarray data, 01:03:08.730 --> 01:03:15.010 align:middle line:84% and you don't have the time or the money to-- 01:03:15.010 --> 01:03:20.040 align:middle line:84% sorry-- to do your own microarrays, you can go to GEO, 01:03:20.040 --> 01:03:23.010 align:middle line:84% and you have 12,000 of them to choose from. 01:03:23.010 --> 01:03:26.160 align:middle line:84% And they also offer a very powerful interface to search-- 01:03:26.160 --> 01:03:28.080 align:middle line:84% since microarray data sets are very large, 01:03:28.080 --> 01:03:30.360 align:middle line:84% they include thousands of measurements. 01:03:30.360 --> 01:03:34.560 align:middle line:90% 01:03:34.560 --> 01:03:37.530 align:middle line:84% They provide a very useful search interface 01:03:37.530 --> 01:03:40.260 align:middle line:84% that allows you to select the data sets we're interested in, 01:03:40.260 --> 01:03:42.060 align:middle line:84% and to extract data from these data sets 01:03:42.060 --> 01:03:45.180 align:middle line:84% and look at, for example, the behavior of the same gene 01:03:45.180 --> 01:03:49.902 align:middle line:84% in different experiments, or different genes 01:03:49.902 --> 01:03:50.860 align:middle line:90% in the same experiment. 01:03:50.860 --> 01:03:53.280 align:middle line:84% And there are lots of different queries 01:03:53.280 --> 01:03:56.160 align:middle line:84% that are common when you work with microarray data. 01:03:56.160 --> 01:03:59.710 align:middle line:90% 01:03:59.710 --> 01:04:01.910 align:middle line:84% The Stanford microarray database, again, 01:04:01.910 --> 01:04:05.987 align:middle line:90% is a repository of all the-- 01:04:05.987 --> 01:04:09.970 align:middle line:84% of a large number of micro experiments performed 01:04:09.970 --> 01:04:11.965 align:middle line:84% at Stanford, and a portion of these are public. 01:04:11.965 --> 01:04:14.990 align:middle line:90% 01:04:14.990 --> 01:04:18.860 align:middle line:84% NCI60, again, from Stanford, is a famous data set 01:04:18.860 --> 01:04:23.610 align:middle line:84% that includes gene expression profiles for 60 human cancer 01:04:23.610 --> 01:04:24.610 align:middle line:90% cell lines. 01:04:24.610 --> 01:04:28.592 align:middle line:84% And the information on drug activity 01:04:28.592 --> 01:04:30.300 align:middle line:84% correlated with gene expression patterns. 01:04:30.300 --> 01:04:34.800 align:middle line:84% So they measure how the gene expression patterns 01:04:34.800 --> 01:04:39.315 align:middle line:84% change when these cell lines are subject to different drugs. 01:04:39.315 --> 01:04:43.700 align:middle line:90% 01:04:43.700 --> 01:04:46.130 align:middle line:84% Other resources for gene expression 01:04:46.130 --> 01:04:49.520 align:middle line:84% are found in different PGA projects, 01:04:49.520 --> 01:04:51.470 align:middle line:84% PGA are programs for genomic applications, 01:04:51.470 --> 01:04:54.350 align:middle line:84% they are are large projects managed by the NIH. 01:04:54.350 --> 01:04:56.280 align:middle line:84% So the [? tracks ?] PGA, for example, 01:04:56.280 --> 01:05:00.890 align:middle line:84% offers 565 microarrays from mouse and rat 01:05:00.890 --> 01:05:03.800 align:middle line:84% models of sleep, infection, hypertension, pulmonary 01:05:03.800 --> 01:05:05.840 align:middle line:90% disease. 01:05:05.840 --> 01:05:09.920 align:middle line:84% The Hopkins PGA, again more than 500 microarrays 01:05:09.920 --> 01:05:13.370 align:middle line:90% from several human diseases. 01:05:13.370 --> 01:05:18.000 align:middle line:84% Cardio genomics provide microarray data 01:05:18.000 --> 01:05:20.630 align:middle line:84% on mouse models of cardiac development and signal 01:05:20.630 --> 01:05:22.280 align:middle line:90% transduction. 01:05:22.280 --> 01:05:24.890 align:middle line:84% And finally, human gene expression index-- 01:05:24.890 --> 01:05:30.780 align:middle line:84% these are just some of the most important most useful 01:05:30.780 --> 01:05:32.430 align:middle line:84% public resources of microarray data. 01:05:32.430 --> 01:05:38.830 align:middle line:90% 01:05:38.830 --> 01:05:45.780 align:middle line:84% OK, I'm going to go through this final part 01:05:45.780 --> 01:05:49.840 align:middle line:84% very quickly because I'm almost out of time. 01:05:49.840 --> 01:05:52.890 align:middle line:84% And if you'd rather stop me with questions, 01:05:52.890 --> 01:05:57.210 align:middle line:84% or if there's anything you would like to discuss about what 01:05:57.210 --> 01:05:58.890 align:middle line:84% I said so far, we could stop here, 01:05:58.890 --> 01:06:04.050 align:middle line:84% or I could just run through this last portion quickly. 01:06:04.050 --> 01:06:07.410 align:middle line:90% 01:06:07.410 --> 01:06:11.420 align:middle line:84% So this last part was about the last step in the process 01:06:11.420 --> 01:06:12.950 align:middle line:90% from proteins to phenotypes. 01:06:12.950 --> 01:06:15.740 align:middle line:90% 01:06:15.740 --> 01:06:18.590 align:middle line:84% I was going to talk about protein databases. 01:06:18.590 --> 01:06:20.210 align:middle line:84% The situation in protein databases 01:06:20.210 --> 01:06:24.500 align:middle line:84% is a bit different from what we've seen so far. 01:06:24.500 --> 01:06:27.890 align:middle line:84% The protein world is much more complex than the DNA and RNA 01:06:27.890 --> 01:06:31.148 align:middle line:84% world for the reasons that I've explained at the beginning. 01:06:31.148 --> 01:06:32.690 align:middle line:84% Some of the reasons are that proteins 01:06:32.690 --> 01:06:34.610 align:middle line:84% interact with each other in very complex ways. 01:06:34.610 --> 01:06:36.800 align:middle line:84% They combine in three dimensions, 01:06:36.800 --> 01:06:38.510 align:middle line:84% they catalyze chemical reactions. 01:06:38.510 --> 01:06:41.990 align:middle line:84% They have a behavioral that is much harder 01:06:41.990 --> 01:06:47.750 align:middle line:84% to describe in [INAUDIBLE] terms than everything else we've 01:06:47.750 --> 01:06:49.740 align:middle line:90% seen so far. 01:06:49.740 --> 01:06:53.000 align:middle line:84% So what protein databases give you is usually 01:06:53.000 --> 01:06:55.460 align:middle line:84% information about the sequence of a protein, 01:06:55.460 --> 01:06:58.710 align:middle line:90% and that's the easy part. 01:06:58.710 --> 01:07:00.860 align:middle line:84% The known or computed three dimensional structure, 01:07:00.860 --> 01:07:03.470 align:middle line:84% the known or inferred function domains. 01:07:03.470 --> 01:07:06.513 align:middle line:84% And ideally, also, the functional protein, 01:07:06.513 --> 01:07:08.430 align:middle line:84% what the protein does in different conditions. 01:07:08.430 --> 01:07:10.040 align:middle line:90% But again, this is-- 01:07:10.040 --> 01:07:13.100 align:middle line:84% we're getting to the area where things 01:07:13.100 --> 01:07:16.850 align:middle line:84% start becoming hard to formalize and to represent 01:07:16.850 --> 01:07:20.640 align:middle line:90% in a computational system. 01:07:20.640 --> 01:07:24.200 align:middle line:84% So as a consequence, protein databases, first of all, 01:07:24.200 --> 01:07:27.830 align:middle line:84% tend to be older, because they were started earlier 01:07:27.830 --> 01:07:29.090 align:middle line:90% than genomic databases. 01:07:29.090 --> 01:07:31.550 align:middle line:84% They are less integrated, they are less complete. 01:07:31.550 --> 01:07:34.970 align:middle line:84% Nomenclature is much less standardized. 01:07:34.970 --> 01:07:38.180 align:middle line:84% So it's harder to work on protein databases 01:07:38.180 --> 01:07:42.230 align:middle line:84% than with all the other resources we've seen so far. 01:07:42.230 --> 01:07:44.990 align:middle line:84% The biggest database is SwissProt, 01:07:44.990 --> 01:07:48.140 align:middle line:84% 120,000 sequenced centuries, 9,000 human proteins 01:07:48.140 --> 01:07:52.010 align:middle line:84% in SwissProt, which is a pretty small number if you think 01:07:52.010 --> 01:07:55.340 align:middle line:84% that we already have complete information in Golden 01:07:55.340 --> 01:07:57.380 align:middle line:84% Path about 20,000 genes, and each gene 01:07:57.380 --> 01:08:02.210 align:middle line:84% is known to, on the average, code for probably two or more 01:08:02.210 --> 01:08:02.820 align:middle line:90% proteins. 01:08:02.820 --> 01:08:06.727 align:middle line:84% So these are the proteins for which we know something, 01:08:06.727 --> 01:08:08.810 align:middle line:84% and they're very, very, very small number compared 01:08:08.810 --> 01:08:11.000 align:middle line:84% to the total number of proteins that 01:08:11.000 --> 01:08:13.820 align:middle line:90% are thought to be in our cells. 01:08:13.820 --> 01:08:18.319 align:middle line:84% It's composed of a core set of data elements, this sequence, 01:08:18.319 --> 01:08:21.410 align:middle line:84% the references, taxonomic data for this protein. 01:08:21.410 --> 01:08:24.830 align:middle line:84% And then our [INAUDIBLE] about the functions of this protein, 01:08:24.830 --> 01:08:27.620 align:middle line:84% domains and sites, the structure, similarities, 01:08:27.620 --> 01:08:32.569 align:middle line:84% association with diseases, variant forms of the protein. 01:08:32.569 --> 01:08:36.680 align:middle line:84% And again, it's hard to link this database with LocusLink 01:08:36.680 --> 01:08:40.399 align:middle line:84% or UniGene, but is its own identifiers for proteins. 01:08:40.399 --> 01:08:44.450 align:middle line:84% But don't need to go into these problems now. 01:08:44.450 --> 01:08:47.750 align:middle line:84% This is a graph that shows you the growth of SwissProt 01:08:47.750 --> 01:08:48.660 align:middle line:90% in recent years. 01:08:48.660 --> 01:08:51.859 align:middle line:84% And as you can see it's growing but at a much smaller rate 01:08:51.859 --> 01:08:54.710 align:middle line:84% than GenBank or other resources like that. 01:08:54.710 --> 01:08:58.189 align:middle line:90% 01:08:58.189 --> 01:09:00.149 align:middle line:84% We have databases about the three 01:09:00.149 --> 01:09:03.410 align:middle line:84% dimensional structural proteins like PDB, 01:09:03.410 --> 01:09:07.010 align:middle line:90% different visualization options. 01:09:07.010 --> 01:09:10.579 align:middle line:84% MMDB is essentially the same thing, but implemented at NCBI. 01:09:10.579 --> 01:09:13.750 align:middle line:90% 01:09:13.750 --> 01:09:17.470 align:middle line:84% PFM at the Sanger Institute is a database of protein domains 01:09:17.470 --> 01:09:18.790 align:middle line:90% and protein families. 01:09:18.790 --> 01:09:22.330 align:middle line:84% They look for domains in the proteins, 01:09:22.330 --> 01:09:23.830 align:middle line:84% and then they look for similarities 01:09:23.830 --> 01:09:26.200 align:middle line:84% between proteins on the basis of the domains that 01:09:26.200 --> 01:09:28.569 align:middle line:90% were identified. 01:09:28.569 --> 01:09:33.189 align:middle line:84% They use similarity measures, they use hidden Markov models. 01:09:33.189 --> 01:09:36.490 align:middle line:84% Again, they have a curated portion 01:09:36.490 --> 01:09:39.479 align:middle line:84% with a small number of protein families. 01:09:39.479 --> 01:09:42.229 align:middle line:84% With the notation there is a high quality. 01:09:42.229 --> 01:09:44.380 align:middle line:84% And then there's a second portion of PFM 01:09:44.380 --> 01:09:49.390 align:middle line:84% that has smaller families of lower quality. 01:09:49.390 --> 01:09:52.200 align:middle line:84% This is an example of a display, a PFM display, 01:09:52.200 --> 01:09:54.930 align:middle line:84% of a protein with all the different domains that 01:09:54.930 --> 01:09:57.270 align:middle line:84% were found in the protein with the tails here. 01:09:57.270 --> 01:10:00.090 align:middle line:84% So give it this nice graphical display. 01:10:00.090 --> 01:10:02.940 align:middle line:90% 01:10:02.940 --> 01:10:05.040 align:middle line:84% I'm going to skip protein interaction databases. 01:10:05.040 --> 01:10:13.350 align:middle line:90% And I want to get to the end. 01:10:13.350 --> 01:10:17.220 align:middle line:84% We're getting to the phenotype and to the spectrum, finally. 01:10:17.220 --> 01:10:18.833 align:middle line:84% And there's just a couple of resources 01:10:18.833 --> 01:10:21.250 align:middle line:84% that have to be cited because they're extremely important. 01:10:21.250 --> 01:10:22.330 align:middle line:90% One of them is OMIM-- 01:10:22.330 --> 01:10:27.090 align:middle line:84% OMIM is a catalog of human genes and genetic disorders. 01:10:27.090 --> 01:10:28.430 align:middle line:90% Again, hosted by the NCBI. 01:10:28.430 --> 01:10:32.010 align:middle line:90% 01:10:32.010 --> 01:10:36.240 align:middle line:84% It's basically a collection of text articles 01:10:36.240 --> 01:10:39.052 align:middle line:84% that talk either about a gene or about a disorder, 01:10:39.052 --> 01:10:40.510 align:middle line:84% and they're linked with each other. 01:10:40.510 --> 01:10:43.660 align:middle line:84% So if you're looking at the entry for a gene, 01:10:43.660 --> 01:10:47.933 align:middle line:84% you can find a description, mechanical feature, 01:10:47.933 --> 01:10:49.350 align:middle line:84% the function mapping, and then you 01:10:49.350 --> 01:10:56.040 align:middle line:84% can find all known correlations between that gene and diseases, 01:10:56.040 --> 01:10:59.010 align:middle line:84% allelic variants, so all known polymorphisms of that gene 01:10:59.010 --> 01:11:02.943 align:middle line:84% with the corresponding clinical outcome if there is any. 01:11:02.943 --> 01:11:04.860 align:middle line:84% And then you can also go the other way around. 01:11:04.860 --> 01:11:08.700 align:middle line:90% 01:11:08.700 --> 01:11:11.430 align:middle line:90% It has 14,000 entries. 01:11:11.430 --> 01:11:15.990 align:middle line:84% Again, these numbers are probably larger by now, 01:11:15.990 --> 01:11:18.473 align:middle line:84% because again this is a graph that shows you how-- 01:11:18.473 --> 01:11:19.890 align:middle line:84% it is not very up to date, but you 01:11:19.890 --> 01:11:23.010 align:middle line:84% can imagine that it's been growing at least at this speed 01:11:23.010 --> 01:11:26.940 align:middle line:90% or faster since '98. 01:11:26.940 --> 01:11:29.520 align:middle line:84% And finally, PubMed, you probably all know 01:11:29.520 --> 01:11:32.218 align:middle line:84% what PubMed is, a database of citations 01:11:32.218 --> 01:11:33.510 align:middle line:90% from the biomedical literature. 01:11:33.510 --> 01:11:36.480 align:middle line:84% It contains 12 million entries starting from 01:11:36.480 --> 01:11:41.550 align:middle line:84% the mid-'60s, and it provides references, abstracts, 01:11:41.550 --> 01:11:45.160 align:middle line:90% linked to online resources. 01:11:45.160 --> 01:11:49.570 align:middle line:84% Full text articles, in some cases, supplementary materials, 01:11:49.570 --> 01:11:53.650 align:middle line:84% and it's one of the most used resources in this field. 01:11:53.650 --> 01:11:56.440 align:middle line:84% They claim they receive 30 million searches per month. 01:11:56.440 --> 01:12:00.390 align:middle line:90% 01:12:00.390 --> 01:12:05.520 align:middle line:84% OK, one last thing-- gene ontology. 01:12:05.520 --> 01:12:07.410 align:middle line:84% Gene ontology is something that stays 01:12:07.410 --> 01:12:09.570 align:middle line:84% at a slightly higher level, above everything 01:12:09.570 --> 01:12:11.670 align:middle line:90% that we've seen so far. 01:12:11.670 --> 01:12:14.400 align:middle line:84% The idea of gene ontology is to build 01:12:14.400 --> 01:12:16.230 align:middle line:84% a dynamic controlled vocabulary that 01:12:16.230 --> 01:12:18.960 align:middle line:84% can be used to describe biological concepts. 01:12:18.960 --> 01:12:21.540 align:middle line:84% If you look at something like OMIM or PubMed, 01:12:21.540 --> 01:12:24.000 align:middle line:84% you're going to find a textual description, for example, 01:12:24.000 --> 01:12:29.792 align:middle line:84% of a disease that references concepts that 01:12:29.792 --> 01:12:32.250 align:middle line:84% need to be precisely defined so that we all know what we're 01:12:32.250 --> 01:12:34.350 align:middle line:84% talking about the same thing when you use the same word. 01:12:34.350 --> 01:12:35.940 align:middle line:84% And the purpose of gene ontology is 01:12:35.940 --> 01:12:41.190 align:middle line:84% to try to do this in at least three domains-- 01:12:41.190 --> 01:12:44.830 align:middle line:84% molecular function, biological process, cellular component. 01:12:44.830 --> 01:12:47.580 align:middle line:84% So it's organized in three taxonomies, 01:12:47.580 --> 01:12:50.970 align:middle line:84% and each taxonomy contains concepts and sub-concepts 01:12:50.970 --> 01:12:56.490 align:middle line:84% and so on, that try to describe everything that 01:12:56.490 --> 01:12:58.200 align:middle line:84% is known about molecular functions, 01:12:58.200 --> 01:13:00.870 align:middle line:84% molecular biological process, and several components 01:13:00.870 --> 01:13:02.790 align:middle line:84% using a standardized nomenclature. 01:13:02.790 --> 01:13:04.720 align:middle line:84% So that when you want to refer, for example, 01:13:04.720 --> 01:13:07.080 align:middle line:84% for to a certain component of a cell, 01:13:07.080 --> 01:13:12.510 align:middle line:84% instead of just saying its name, you can cite the gene ontology 01:13:12.510 --> 01:13:14.640 align:middle line:84% term that describes that component, 01:13:14.640 --> 01:13:17.212 align:middle line:84% and everybody else will be able to go to gene ontology 01:13:17.212 --> 01:13:19.170 align:middle line:84% and see what's the exact definition of the word 01:13:19.170 --> 01:13:21.360 align:middle line:90% you're using. 01:13:21.360 --> 01:13:23.190 align:middle line:90% It's a work in progress-- 01:13:23.190 --> 01:13:25.260 align:middle line:84% still very far from being complete. 01:13:25.260 --> 01:13:26.907 align:middle line:84% It has all the usual problems that 01:13:26.907 --> 01:13:28.740 align:middle line:84% occur when you're trying to build taxonomies 01:13:28.740 --> 01:13:32.295 align:middle line:84% that it's very hard to formalize things that 01:13:32.295 --> 01:13:33.420 align:middle line:90% come from natural language. 01:13:33.420 --> 01:13:36.120 align:middle line:84% So it could find exact definition of all the terms 01:13:36.120 --> 01:13:38.710 align:middle line:84% that people use, especially in this field is very hard. 01:13:38.710 --> 01:13:41.550 align:middle line:84% But this is where they are now, and it's a work in progress, 01:13:41.550 --> 01:13:46.380 align:middle line:84% so it will keep growing in the future. 01:13:46.380 --> 01:13:50.430 align:middle line:84% And this is a view of taxonomy, for example, 01:13:50.430 --> 01:13:53.060 align:middle line:90% for biological process. 01:13:53.060 --> 01:13:55.130 align:middle line:84% If you are talking about site communication, 01:13:55.130 --> 01:13:58.190 align:middle line:84% then response to external stimulus 01:13:58.190 --> 01:14:00.920 align:middle line:90% is a subclass of communication. 01:14:00.920 --> 01:14:03.320 align:middle line:84% Response the biology stimulus, the first response 01:14:03.320 --> 01:14:05.210 align:middle line:90% is a subclass again of all this. 01:14:05.210 --> 01:14:09.440 align:middle line:84% And if you want to talk about the immune response, 01:14:09.440 --> 01:14:11.900 align:middle line:84% you can cite this biology term, and everybody 01:14:11.900 --> 01:14:14.660 align:middle line:84% will be able to go to gene ontology 01:14:14.660 --> 01:14:18.800 align:middle line:84% and see exactly where this term is in the taxonomy of concepts 01:14:18.800 --> 01:14:21.900 align:middle line:90% about biological processes. 01:14:21.900 --> 01:14:26.570 align:middle line:90% OK, I think we're out of time. 01:14:26.570 --> 01:14:29.090 align:middle line:84% Well, just a conclusion slide that I'll just 01:14:29.090 --> 01:14:35.270 align:middle line:84% let you read, because I think it's just repeating what we're 01:14:35.270 --> 01:14:41.310 align:middle line:84% saying so far that we are drowning in data 01:14:41.310 --> 01:14:43.600 align:middle line:84% and converting this data into knowledge is not easy. 01:14:43.600 --> 01:14:47.970 align:middle line:84% We need automated tools to access this data, 01:14:47.970 --> 01:14:52.480 align:middle line:84% to make sense of it, to convert into formats that we can use. 01:14:52.480 --> 01:14:56.520 align:middle line:84% And of course, this is a challenging task, 01:14:56.520 --> 01:14:58.890 align:middle line:84% because as we saw, biomedical data 01:14:58.890 --> 01:15:01.230 align:middle line:84% covers the whole spectrum of knowledge representation 01:15:01.230 --> 01:15:05.000 align:middle line:84% and management techniques that we know about. 01:15:05.000 --> 01:15:06.000 align:middle line:90%