WEBVTT 00:00:00.612 --> 00:00:03.374 For the next 16 minutes, I'm going to take you on a journey 00:00:03.398 --> 00:00:06.484 that is probably the biggest dream of humanity: 00:00:06.508 --> 00:00:08.523 to understand the code of life. NOTE Paragraph 00:00:09.072 --> 00:00:11.815 So for me, everything started many, many years ago 00:00:11.839 --> 00:00:14.562 when I met the first 3D printer. 00:00:14.586 --> 00:00:16.260 The concept was fascinating. 00:00:16.284 --> 00:00:18.306 A 3D printer needs three elements: 00:00:18.330 --> 00:00:22.464 a bit of information, some raw material, some energy, 00:00:22.488 --> 00:00:25.822 and it can produce any object that was not there before. NOTE Paragraph 00:00:26.517 --> 00:00:28.654 I was doing physics, I was coming back home 00:00:28.678 --> 00:00:32.116 and I realized that I actually always knew a 3D printer. 00:00:32.140 --> 00:00:33.476 And everyone does. 00:00:33.500 --> 00:00:34.658 It was my mom. NOTE Paragraph 00:00:34.682 --> 00:00:35.683 (Laughter) NOTE Paragraph 00:00:35.707 --> 00:00:38.121 My mom takes three elements: 00:00:38.145 --> 00:00:42.118 a bit of information, which is between my father and my mom in this case, 00:00:42.142 --> 00:00:46.299 raw elements and energy in the same media, that is food, 00:00:46.323 --> 00:00:48.831 and after several months, produces me. 00:00:48.855 --> 00:00:50.667 And I was not existent before. NOTE Paragraph 00:00:50.691 --> 00:00:54.453 So apart from the shock of my mom discovering that she was a 3D printer, 00:00:54.477 --> 00:00:59.215 I immediately got mesmerized by that piece, 00:00:59.239 --> 00:01:00.956 the first one, the information. 00:01:00.980 --> 00:01:03.231 What amount of information does it take 00:01:03.255 --> 00:01:05.191 to build and assemble a human? 00:01:05.215 --> 00:01:06.789 Is it much? Is it little? 00:01:06.813 --> 00:01:08.993 How many thumb drives can you fill? NOTE Paragraph 00:01:09.017 --> 00:01:11.641 Well, I was studying physics at the beginning 00:01:11.665 --> 00:01:17.262 and I took this approximation of a human as a gigantic Lego piece. 00:01:17.286 --> 00:01:21.071 So, imagine that the building blocks are little atoms 00:01:21.095 --> 00:01:25.748 and there is a hydrogen here, a carbon here, a nitrogen here. 00:01:25.772 --> 00:01:27.343 So in the first approximation, 00:01:27.367 --> 00:01:31.710 if I can list the number of atoms that compose a human being, 00:01:31.734 --> 00:01:33.121 I can build it. 00:01:33.145 --> 00:01:35.174 Now, you can run some numbers 00:01:35.198 --> 00:01:38.475 and that happens to be quite an astonishing number. 00:01:38.499 --> 00:01:41.256 So the number of atoms, 00:01:41.280 --> 00:01:46.035 the file that I will save in my thumb drive to assemble a little baby, 00:01:46.059 --> 00:01:50.726 will actually fill an entire Titanic of thumb drives -- 00:01:50.750 --> 00:01:53.468 multiplied 2,000 times. 00:01:53.957 --> 00:01:57.358 This is the miracle of life. 00:01:57.382 --> 00:01:59.994 Every time you see from now on a pregnant lady, 00:02:00.018 --> 00:02:02.874 she's assembling the biggest amount of information 00:02:02.898 --> 00:02:04.454 that you will ever encounter. 00:02:04.478 --> 00:02:07.428 Forget big data, forget anything you heard of. 00:02:07.452 --> 00:02:10.333 This is the biggest amount of information that exists. NOTE Paragraph 00:02:10.357 --> 00:02:14.190 (Applause) NOTE Paragraph 00:02:14.214 --> 00:02:18.858 But nature, fortunately, is much smarter than a young physicist, 00:02:18.882 --> 00:02:22.458 and in four billion years, managed to pack this information 00:02:22.482 --> 00:02:25.187 in a small crystal we call DNA. 00:02:25.605 --> 00:02:29.917 We met it for the first time in 1950 when Rosalind Franklin, 00:02:29.941 --> 00:02:31.497 an amazing scientist, a woman, 00:02:31.521 --> 00:02:32.910 took a picture of it. 00:02:32.934 --> 00:02:38.122 But it took us more than 40 years to finally poke inside a human cell, 00:02:38.146 --> 00:02:39.748 take out this crystal, 00:02:39.772 --> 00:02:42.852 unroll it, and read it for the first time. 00:02:43.615 --> 00:02:46.856 The code comes out to be a fairly simple alphabet, 00:02:46.880 --> 00:02:50.652 four letters: A, T, C and G. 00:02:50.676 --> 00:02:54.166 And to build a human, you need three billion of them. 00:02:54.933 --> 00:02:56.112 Three billion. 00:02:56.136 --> 00:02:57.715 How many are three billion? 00:02:57.739 --> 00:03:00.501 It doesn't really make any sense as a number, right? NOTE Paragraph 00:03:00.525 --> 00:03:04.610 So I was thinking how I could explain myself better 00:03:04.634 --> 00:03:07.684 about how big and enormous this code is. 00:03:07.708 --> 00:03:10.762 But there is -- I mean, I'm going to have some help, 00:03:10.786 --> 00:03:14.013 and the best person to help me introduce the code 00:03:14.037 --> 00:03:17.559 is actually the first man to sequence it, Dr. Craig Venter. 00:03:17.583 --> 00:03:20.973 So welcome onstage, Dr. Craig Venter. NOTE Paragraph 00:03:20.997 --> 00:03:27.928 (Applause) NOTE Paragraph 00:03:27.952 --> 00:03:30.208 Not the man in the flesh, 00:03:31.448 --> 00:03:33.793 but for the first time in history, 00:03:33.817 --> 00:03:37.279 this is the genome of a specific human, 00:03:37.303 --> 00:03:41.063 printed page-by-page, letter-by-letter: 00:03:41.087 --> 00:03:45.083 262,000 pages of information, 00:03:45.107 --> 00:03:49.471 450 kilograms, shipped from the United States to Canada 00:03:49.495 --> 00:03:54.338 thanks to Bruno Bowden, Lulu.com, a start-up, did everything. 00:03:54.362 --> 00:03:55.825 It was an amazing feat. NOTE Paragraph 00:03:55.849 --> 00:04:00.146 But this is the visual perception of what is the code of life. 00:04:00.170 --> 00:04:02.648 And now, for the first time, I can do something fun. 00:04:02.672 --> 00:04:05.219 I can actually poke inside it and read. 00:04:05.243 --> 00:04:09.868 So let me take an interesting book ... like this one. 00:04:13.077 --> 00:04:15.611 I have an annotation; it's a fairly big book. 00:04:15.635 --> 00:04:19.362 So just to let you see what is the code of life. 00:04:20.566 --> 00:04:23.957 Thousands and thousands and thousands 00:04:23.981 --> 00:04:26.651 and millions of letters. 00:04:26.675 --> 00:04:29.071 And they apparently make sense. 00:04:29.095 --> 00:04:30.852 Let's get to a specific part. 00:04:31.571 --> 00:04:32.933 Let me read it to you: NOTE Paragraph 00:04:32.957 --> 00:04:33.978 (Laughter) NOTE Paragraph 00:04:34.002 --> 00:04:38.008 "AAG, AAT, ATA." NOTE Paragraph 00:04:38.965 --> 00:04:41.032 To you it sounds like mute letters, 00:04:41.056 --> 00:04:45.097 but this sequence gives the color of the eyes to Craig. 00:04:45.633 --> 00:04:47.565 I'll show you another part of the book. 00:04:47.589 --> 00:04:49.683 This is actually a little more complicated. NOTE Paragraph 00:04:50.983 --> 00:04:53.630 Chromosome 14, book 132: NOTE Paragraph 00:04:53.654 --> 00:04:55.744 (Laughter) NOTE Paragraph 00:04:55.768 --> 00:04:57.045 As you might expect. NOTE Paragraph 00:04:57.069 --> 00:05:00.535 (Laughter) NOTE Paragraph 00:05:02.857 --> 00:05:07.364 "ATT, CTT, GATT." NOTE Paragraph 00:05:08.329 --> 00:05:10.016 This human is lucky, 00:05:10.040 --> 00:05:14.557 because if you miss just two letters in this position -- 00:05:14.581 --> 00:05:16.458 two letters of our three billion -- 00:05:16.482 --> 00:05:18.501 he will be condemned to a terrible disease: 00:05:18.525 --> 00:05:19.965 cystic fibrosis. 00:05:19.989 --> 00:05:23.402 We have no cure for it, we don't know how to solve it, 00:05:23.426 --> 00:05:27.181 and it's just two letters of difference from what we are. NOTE Paragraph 00:05:27.585 --> 00:05:30.290 A wonderful book, a mighty book, 00:05:31.115 --> 00:05:33.113 a mighty book that helped me understand 00:05:33.137 --> 00:05:35.890 and show you something quite remarkable. 00:05:36.480 --> 00:05:40.915 Every one of you -- what makes me, me and you, you -- 00:05:40.939 --> 00:05:43.893 is just about five million of these, 00:05:43.917 --> 00:05:45.145 half a book. 00:05:46.015 --> 00:05:47.678 For the rest, 00:05:47.702 --> 00:05:50.264 we are all absolutely identical. 00:05:51.008 --> 00:05:55.026 Five hundred pages is the miracle of life that you are. 00:05:55.050 --> 00:05:57.581 The rest, we all share it. 00:05:57.605 --> 00:06:00.514 So think about that again when we think that we are different. 00:06:00.538 --> 00:06:02.759 This is the amount that we share. NOTE Paragraph 00:06:03.441 --> 00:06:06.870 So now that I have your attention, 00:06:06.894 --> 00:06:08.253 the next question is: 00:06:08.277 --> 00:06:09.428 How do I read it? 00:06:09.452 --> 00:06:10.961 How do I make sense out of it? 00:06:11.409 --> 00:06:15.649 Well, for however good you can be at assembling Swedish furniture, 00:06:15.673 --> 00:06:19.236 this instruction manual is nothing you can crack in your life. NOTE Paragraph 00:06:19.260 --> 00:06:20.863 (Laughter) NOTE Paragraph 00:06:20.887 --> 00:06:23.999 And so, in 2014, two famous TEDsters, 00:06:24.023 --> 00:06:26.563 Peter Diamandis and Craig Venter himself, 00:06:26.587 --> 00:06:28.514 decided to assemble a new company. 00:06:28.538 --> 00:06:29.950 Human Longevity was born, 00:06:29.974 --> 00:06:31.344 with one mission: 00:06:31.368 --> 00:06:33.229 trying everything we can try 00:06:33.253 --> 00:06:36.012 and learning everything we can learn from these books, 00:06:36.036 --> 00:06:37.741 with one target -- 00:06:38.862 --> 00:06:41.663 making real the dream of personalized medicine, 00:06:41.687 --> 00:06:45.454 understanding what things should be done to have better health 00:06:45.478 --> 00:06:47.761 and what are the secrets in these books. NOTE Paragraph 00:06:48.329 --> 00:06:52.579 An amazing team, 40 data scientists and many, many more people, 00:06:52.603 --> 00:06:53.953 a pleasure to work with. 00:06:53.977 --> 00:06:56.230 The concept is actually very simple. 00:06:56.254 --> 00:06:59.412 We're going to use a technology called machine learning. 00:06:59.436 --> 00:07:03.975 On one side, we have genomes -- thousands of them. 00:07:03.999 --> 00:07:07.996 On the other side, we collected the biggest database of human beings: 00:07:08.020 --> 00:07:12.316 phenotypes, 3D scan, NMR -- everything you can think of. 00:07:12.340 --> 00:07:15.239 Inside there, on these two opposite sides, 00:07:15.263 --> 00:07:17.705 there is the secret of translation. 00:07:17.729 --> 00:07:20.201 And in the middle, we build a machine. 00:07:20.801 --> 00:07:23.186 We build a machine and we train a machine -- 00:07:23.210 --> 00:07:26.420 well, not exactly one machine, many, many machines -- 00:07:26.444 --> 00:07:30.988 to try to understand and translate the genome in a phenotype. 00:07:31.362 --> 00:07:34.702 What are those letters, and what do they do? 00:07:34.726 --> 00:07:37.473 It's an approach that can be used for everything, 00:07:37.497 --> 00:07:40.490 but using it in genomics is particularly complicated. 00:07:40.514 --> 00:07:43.790 Little by little we grew and we wanted to build different challenges. 00:07:43.814 --> 00:07:46.546 We started from the beginning, from common traits. 00:07:46.570 --> 00:07:49.173 Common traits are comfortable because they are common, 00:07:49.197 --> 00:07:50.381 everyone has them. NOTE Paragraph 00:07:50.405 --> 00:07:52.899 So we started to ask our questions: 00:07:52.923 --> 00:07:54.303 Can we predict height? 00:07:54.985 --> 00:07:57.162 Can we read the books and predict your height? 00:07:57.186 --> 00:07:58.337 Well, we actually can, 00:07:58.361 --> 00:08:00.154 with five centimeters of precision. 00:08:00.178 --> 00:08:03.313 BMI is fairly connected to your lifestyle, 00:08:03.337 --> 00:08:07.201 but we still can, we get in the ballpark, eight kilograms of precision. 00:08:07.225 --> 00:08:08.456 Can we predict eye color? 00:08:08.480 --> 00:08:09.638 Yeah, we can. 00:08:09.662 --> 00:08:10.986 Eighty percent accuracy. 00:08:11.466 --> 00:08:13.324 Can we predict skin color? 00:08:13.348 --> 00:08:15.789 Yeah we can, 80 percent accuracy. 00:08:15.813 --> 00:08:17.153 Can we predict age? 00:08:18.121 --> 00:08:21.860 We can, because apparently, the code changes during your life. 00:08:21.884 --> 00:08:25.166 It gets shorter, you lose pieces, it gets insertions. 00:08:25.190 --> 00:08:27.745 We read the signals, and we make a model. NOTE Paragraph 00:08:28.438 --> 00:08:29.913 Now, an interesting challenge: 00:08:29.937 --> 00:08:31.666 Can we predict a human face? 00:08:33.014 --> 00:08:34.292 It's a little complicated, 00:08:34.316 --> 00:08:37.507 because a human face is scattered among millions of these letters. 00:08:37.531 --> 00:08:40.160 And a human face is not a very well-defined object. 00:08:40.184 --> 00:08:42.235 So, we had to build an entire tier of it 00:08:42.259 --> 00:08:44.969 to learn and teach a machine what a face is, 00:08:44.993 --> 00:08:47.030 and embed and compress it. 00:08:47.054 --> 00:08:49.302 And if you're comfortable with machine learning, 00:08:49.326 --> 00:08:51.610 you understand what the challenge is here. NOTE Paragraph 00:08:52.108 --> 00:08:58.099 Now, after 15 years -- 15 years after we read the first sequence -- 00:08:58.123 --> 00:09:01.025 this October, we started to see some signals. 00:09:01.049 --> 00:09:03.504 And it was a very emotional moment. 00:09:03.528 --> 00:09:07.273 What you see here is a subject coming in our lab. 00:09:07.619 --> 00:09:09.547 This is a face for us. 00:09:09.571 --> 00:09:13.202 So we take the real face of a subject, we reduce the complexity, 00:09:13.226 --> 00:09:15.196 because not everything is in your face -- 00:09:15.220 --> 00:09:19.006 lots of features and defects and asymmetries come from your life. 00:09:19.030 --> 00:09:22.499 We symmetrize the face, and we run our algorithm. 00:09:23.245 --> 00:09:25.143 The results that I show you right now, 00:09:25.167 --> 00:09:28.539 this is the prediction we have from the blood. NOTE Paragraph 00:09:29.596 --> 00:09:31.120 (Applause) NOTE Paragraph 00:09:31.144 --> 00:09:32.579 Wait a second. 00:09:32.603 --> 00:09:37.295 In these seconds, your eyes are watching, left and right, left and right, 00:09:37.319 --> 00:09:41.249 and your brain wants those pictures to be identical. 00:09:41.273 --> 00:09:43.719 So I ask you to do another exercise, to be honest. 00:09:43.743 --> 00:09:46.030 Please search for the differences, 00:09:46.054 --> 00:09:47.415 which are many. 00:09:47.439 --> 00:09:50.042 The biggest amount of signal comes from gender, 00:09:50.066 --> 00:09:55.267 then there is age, BMI, the ethnicity component of a human. 00:09:55.291 --> 00:09:59.002 And scaling up over that signal is much more complicated. 00:09:59.026 --> 00:10:02.276 But what you see here, even in the differences, 00:10:02.300 --> 00:10:05.895 lets you understand that we are in the right ballpark, 00:10:05.919 --> 00:10:07.267 that we are getting closer. 00:10:07.291 --> 00:10:09.640 And it's already giving you some emotions. NOTE Paragraph 00:10:09.664 --> 00:10:12.367 This is another subject that comes in place, 00:10:12.391 --> 00:10:13.800 and this is a prediction. 00:10:13.824 --> 00:10:18.420 A little smaller face, we didn't get the complete cranial structure, 00:10:18.444 --> 00:10:21.095 but still, it's in the ballpark. 00:10:21.634 --> 00:10:23.858 This is a subject that comes in our lab, 00:10:23.882 --> 00:10:25.325 and this is the prediction. 00:10:26.056 --> 00:10:30.732 So these people have never been seen in the training of the machine. 00:10:30.756 --> 00:10:33.593 These are the so-called "held-out" set. 00:10:33.617 --> 00:10:37.357 But these are people that you will probably never believe. 00:10:37.381 --> 00:10:40.057 We're publishing everything in a scientific publication, 00:10:40.081 --> 00:10:41.232 you can read it. NOTE Paragraph 00:10:41.256 --> 00:10:43.600 But since we are onstage, Chris challenged me. 00:10:43.624 --> 00:10:47.250 I probably exposed myself and tried to predict 00:10:47.274 --> 00:10:50.105 someone that you might recognize. 00:10:50.470 --> 00:10:54.895 So, in this vial of blood -- and believe me, you have no idea 00:10:54.919 --> 00:10:57.799 what we had to do to have this blood now, here -- 00:10:57.823 --> 00:11:01.724 in this vial of blood is the amount of biological information 00:11:01.748 --> 00:11:04.025 that we need to do a full genome sequence. 00:11:04.049 --> 00:11:06.119 We just need this amount. 00:11:06.528 --> 00:11:09.733 We ran this sequence, and I'm going to do it with you. 00:11:09.757 --> 00:11:13.736 And we start to layer up all the understanding we have. 00:11:13.760 --> 00:11:17.110 In the vial of blood, we predicted he's a male. 00:11:17.134 --> 00:11:18.498 And the subject is a male. 00:11:18.996 --> 00:11:21.434 We predict that he's a meter and 76 cm. 00:11:21.458 --> 00:11:23.850 The subject is a meter and 77 cm. 00:11:23.874 --> 00:11:27.984 So, we predicted that he's 76; the subject is 82. 00:11:28.701 --> 00:11:31.333 We predict his age, 38. 00:11:31.357 --> 00:11:33.261 The subject is 35. 00:11:33.851 --> 00:11:35.975 We predict his eye color. 00:11:36.824 --> 00:11:38.035 Too dark. 00:11:38.059 --> 00:11:39.614 We predict his skin color. 00:11:40.026 --> 00:11:41.436 We are almost there. 00:11:41.899 --> 00:11:43.272 That's his face. NOTE Paragraph 00:11:45.172 --> 00:11:48.441 Now, the reveal moment: 00:11:48.465 --> 00:11:50.235 the subject is this person. NOTE Paragraph 00:11:50.259 --> 00:11:52.194 (Laughter) NOTE Paragraph 00:11:52.218 --> 00:11:54.276 And I did it intentionally. 00:11:54.300 --> 00:11:57.992 I am a very particular and peculiar ethnicity. 00:11:58.016 --> 00:12:00.966 Southern European, Italians -- they never fit in models. 00:12:00.990 --> 00:12:06.120 And it's particular -- that ethnicity is a complex corner case for our model. 00:12:06.144 --> 00:12:07.653 But there is another point. 00:12:07.677 --> 00:12:11.154 So, one of the things that we use a lot to recognize people 00:12:11.178 --> 00:12:12.900 will never be written in the genome. 00:12:12.924 --> 00:12:15.241 It's our free will, it's how I look. 00:12:15.265 --> 00:12:18.494 Not my haircut in this case, but my beard cut. 00:12:18.518 --> 00:12:22.071 So I'm going to show you, I'm going to, in this case, transfer it -- 00:12:22.095 --> 00:12:24.860 and this is nothing more than Photoshop, no modeling -- 00:12:24.884 --> 00:12:26.597 the beard on the subject. 00:12:26.621 --> 00:12:30.093 And immediately, we get much, much better in the feeling. NOTE Paragraph 00:12:30.955 --> 00:12:33.664 So, why do we do this? 00:12:35.938 --> 00:12:41.078 We certainly don't do it for predicting height 00:12:41.102 --> 00:12:43.474 or taking a beautiful picture out of your blood. 00:12:44.390 --> 00:12:48.408 We do it because the same technology and the same approach, 00:12:48.432 --> 00:12:50.952 the machine learning of this code, 00:12:50.976 --> 00:12:54.113 is helping us to understand how we work, 00:12:54.137 --> 00:12:55.623 how your body works, 00:12:55.647 --> 00:12:57.312 how your body ages, 00:12:57.336 --> 00:13:00.105 how disease generates in your body, 00:13:00.129 --> 00:13:03.101 how your cancer grows and develops, 00:13:03.125 --> 00:13:04.908 how drugs work 00:13:04.932 --> 00:13:07.246 and if they work on your body. NOTE Paragraph 00:13:07.713 --> 00:13:09.380 This is a huge challenge. 00:13:09.894 --> 00:13:11.532 This is a challenge that we share 00:13:11.556 --> 00:13:14.135 with thousands of other researchers around the world. 00:13:14.159 --> 00:13:16.381 It's called personalized medicine. 00:13:17.125 --> 00:13:20.585 It's the ability to move from a statistical approach 00:13:20.609 --> 00:13:22.641 where you're a dot in the ocean, 00:13:22.665 --> 00:13:24.478 to a personalized approach, 00:13:24.502 --> 00:13:26.687 where we read all these books 00:13:26.711 --> 00:13:29.575 and we get an understanding of exactly how you are. 00:13:30.260 --> 00:13:33.622 But it is a particularly complicated challenge, 00:13:33.646 --> 00:13:37.644 because of all these books, as of today, 00:13:37.668 --> 00:13:40.310 we just know probably two percent: 00:13:41.027 --> 00:13:44.680 four books of more than 175. NOTE Paragraph 00:13:46.021 --> 00:13:49.227 And this is not the topic of my talk, 00:13:50.145 --> 00:13:52.743 because we will learn more. 00:13:53.378 --> 00:13:56.047 There are the best minds in the world on this topic. 00:13:57.048 --> 00:13:58.882 The prediction will get better, 00:13:58.906 --> 00:14:01.159 the model will get more precise. 00:14:01.183 --> 00:14:03.041 And the more we learn, 00:14:03.065 --> 00:14:07.895 the more we will be confronted with decisions 00:14:07.919 --> 00:14:10.940 that we never had to face before 00:14:10.964 --> 00:14:12.399 about life, 00:14:12.423 --> 00:14:14.097 about death, 00:14:14.121 --> 00:14:15.724 about parenting. NOTE Paragraph 00:14:20.626 --> 00:14:25.372 So, we are touching the very inner detail on how life works. 00:14:26.118 --> 00:14:29.276 And it's a revolution that cannot be confined 00:14:29.300 --> 00:14:31.959 in the domain of science or technology. 00:14:32.960 --> 00:14:35.204 This must be a global conversation. 00:14:35.798 --> 00:14:41.015 We must start to think of the future we're building as a humanity. 00:14:41.039 --> 00:14:45.103 We need to interact with creatives, with artists, with philosophers, 00:14:45.127 --> 00:14:46.637 with politicians. 00:14:46.661 --> 00:14:47.819 Everyone is involved, 00:14:47.843 --> 00:14:50.668 because it's the future of our species. 00:14:51.273 --> 00:14:55.241 Without fear, but with the understanding 00:14:55.265 --> 00:14:59.136 that the decisions that we make in the next year 00:14:59.160 --> 00:15:02.949 will change the course of history forever. NOTE Paragraph 00:15:03.732 --> 00:15:04.892 Thank you. NOTE Paragraph 00:15:04.916 --> 00:15:15.075 (Applause)