1 00:00:10,426 --> 00:00:12,296 >> Sally: Hey Dude, how are you? 2 00:00:13,056 --> 00:00:14,606 >> Dude: I'm not doing too well. 3 00:00:14,886 --> 00:00:18,246 I'm trying to figure which device to use to turn off the gas-o-matic module 4 00:00:18,496 --> 00:00:20,426 and stop Buddy from growing too big. 5 00:00:20,856 --> 00:00:22,486 But I don't know how to choose! 6 00:00:23,256 --> 00:00:27,986 >> Sally: Hmm, well choosing the right device for your system can be pretty hard. 7 00:00:28,466 --> 00:00:29,766 Why don't we work on this together? 8 00:00:30,686 --> 00:00:32,106 >> Dude: That would be great. 9 00:00:32,506 --> 00:00:34,516 I am just going around in circles here. 10 00:00:35,366 --> 00:00:38,346 >> Sally: Well I'd suggest that you use a digital device 11 00:00:38,346 --> 00:00:42,176 to turn off the gas-o-matic module when Buddy got to a certain size. 12 00:00:42,696 --> 00:00:43,106 >> Dude: What's that? 13 00:00:43,946 --> 00:00:49,616 >> Sally: Well let's start with the basics, all digital devices have inputs and outputs 14 00:00:49,726 --> 00:00:54,136 that are always in one of two states: either ON or OFF. 15 00:00:54,946 --> 00:00:59,116 People who work with electrical devices sometimes call the two states HIGH 16 00:00:59,116 --> 00:01:03,686 and LOW because of the high or low voltages that the devices receive or produce. 17 00:01:04,586 --> 00:01:09,626 Other people think of the two states as true/false or as the numbers 1 and 0. 18 00:01:10,136 --> 00:01:11,046 It really doesn't matter. 19 00:01:11,546 --> 00:01:15,186 What's important for us is that if you hook up the output of a digital device 20 00:01:15,256 --> 00:01:19,936 to the gas-o-matic, when the device produces a high signal, Buddy will inflate. 21 00:01:20,226 --> 00:01:23,946 When the device produces a low signal, Buddy will stop inflating 22 00:01:24,016 --> 00:01:26,186 and he'll just stay the same size for a while. 23 00:01:26,596 --> 00:01:28,806 >> Dude: That's cool but why make just two states? 24 00:01:29,036 --> 00:01:32,726 Wouldn't it be cooler to make buddy get bigger really fast at first and then slowed 25 00:01:32,726 --> 00:01:34,836 down when he's close to full size? 26 00:01:34,836 --> 00:01:37,596 I really don't want to see him growing out of control like before. 27 00:01:38,016 --> 00:01:40,666 Couldn't we build an even more interesting system from devices 28 00:01:40,666 --> 00:01:43,166 with more states, like three or even ten? 29 00:01:43,816 --> 00:01:48,766 >> Sally: Those are all really interesting ideas but here's why two states makes the most sense. 30 00:01:49,426 --> 00:01:54,246 Let's imagine that you're measuring the output signal from your digital device over time. 31 00:01:55,076 --> 00:01:59,196 Depending on what the input to the digital device is, the output will switch 32 00:01:59,196 --> 00:02:02,636 between its two possible states ON and OFF. 33 00:02:03,176 --> 00:02:06,826 But of course, this drawing is actually a bit misleading. 34 00:02:07,316 --> 00:02:12,076 See all digital devices, whether they are genetic devices or electrical devices, 35 00:02:12,196 --> 00:02:17,356 never produce a signal that is perfectly ON or perfectly OFF like I've drawn here. 36 00:02:17,776 --> 00:02:20,546 There are always some minor fluctuations in the signal. 37 00:02:21,476 --> 00:02:24,406 These fluctuations aren't real changes in state of course. 38 00:02:24,716 --> 00:02:29,336 But sometimes, if the fluctuations are too big, then they might tell the gas-o-matic 39 00:02:29,336 --> 00:02:31,796 to switch on when you really want it off. 40 00:02:32,516 --> 00:02:33,406 >> Dude: Oh no. 41 00:02:33,406 --> 00:02:34,756 I definitely don't want that. 42 00:02:34,916 --> 00:02:37,566 It took forever to clean up the lab last time that happened. 43 00:02:38,246 --> 00:02:42,886 So the reason we limit the devices to two states is because they work better when there's noise? 44 00:02:43,376 --> 00:02:44,216 >> Sally: Exactly! 45 00:02:45,006 --> 00:02:46,986 >> Dude: And you think Buddy's growth is noisy? 46 00:02:47,256 --> 00:02:47,996 >> Sally: Definitely! 47 00:02:47,996 --> 00:02:50,386 He changes a little just when he breathes. 48 00:02:50,506 --> 00:02:54,966 And he can grow and shrink a lot depending on how much food he has and what's around him. 49 00:02:55,776 --> 00:02:59,056 And sometimes I think he changes size just to show off... 50 00:02:59,686 --> 00:03:02,156 >> Dude: So it sounds like digital devices are the way to go! 51 00:03:02,876 --> 00:03:07,236 It will tell the gas-o-matic device to turn on if Buddy's really growing and turn off 52 00:03:07,236 --> 00:03:11,326 if he's full size, but ignore those in between size fluctuations he likes 53 00:03:11,326 --> 00:03:15,966 to make...but how big are the fluctuations that the device will ignore? 54 00:03:16,456 --> 00:03:17,736 How does it know? 55 00:03:18,826 --> 00:03:21,836 >> Sally: Well the device only knows what we tell it, 56 00:03:21,836 --> 00:03:24,246 and we can split up ON and OFF the way we choose. 57 00:03:24,976 --> 00:03:28,016 >> Dude: OK then...I'll put the ON/OFF border here! 58 00:03:29,026 --> 00:03:30,746 >> Sally: Dude, weren't you listening! 59 00:03:31,076 --> 00:03:35,616 If the signal is somewhere near this boundary and noisy, then the device will from switch 60 00:03:35,746 --> 00:03:39,396 ON to OFF or OFF to ON when you don't want it to. 61 00:03:39,396 --> 00:03:40,136 Here's what we can do. 62 00:03:40,296 --> 00:03:43,366 Let's call just this small part the "valid" signals 63 00:03:43,526 --> 00:03:47,026 and leave all this in between as "invalid" signals. 64 00:03:47,226 --> 00:03:52,606 That way the device won't confuse ON signal values with OFF signal values so easily. 65 00:03:53,386 --> 00:03:54,516 >> Dude: Aw Sally. 66 00:03:54,856 --> 00:03:56,296 Maybe I'm not getting this at all. 67 00:03:56,746 --> 00:04:00,596 It looks like you just put in two boundaries and now you've got twice as much trouble. 68 00:04:01,076 --> 00:04:06,026 What's to stop the noise from switching between valid and invalid signals here AND here? 69 00:04:06,286 --> 00:04:09,356 >> Sally: Dude, I actually think you've got the hang of this. 70 00:04:09,806 --> 00:04:15,576 Here's the trick that engineers use: make the valid output ranges smaller than the input! 71 00:04:15,826 --> 00:04:20,036 This way the quality of the output signal is always better than the quality 72 00:04:20,036 --> 00:04:25,426 of the input signal and even if there is noise on the input, the digital device will get rid 73 00:04:25,426 --> 00:04:28,036 of it but producing a better output signal. 74 00:04:28,656 --> 00:04:33,196 This trick even has a name: Engineers call the difference between the valid input 75 00:04:33,196 --> 00:04:35,566 and output ranges the noise margin. 76 00:04:36,316 --> 00:04:37,196 >> Dude: Noise margins. 77 00:04:37,516 --> 00:04:39,016 I like it! 78 00:04:39,516 --> 00:04:41,736 Digital devices sound wicked useful. 79 00:04:41,946 --> 00:04:43,856 But can we really build them out of DNA? 80 00:04:44,776 --> 00:04:47,726 >> Sally: Some have already been built I can show you how they work 81 00:04:47,726 --> 00:04:49,596 in some cells I've got growing in the lab. 82 00:04:50,066 --> 00:04:53,576 The DNA device you want may be there already and ready to go. 83 00:04:54,216 --> 00:04:56,146 Do you have some time to come to the lab now? 84 00:04:56,696 --> 00:04:58,356 >> Dude: Can I answer in digital?