1
00:00:15,356 --> 00:00:28,996
Speaker 1: Pushkin. I'm Jacob Goldstein and this is What's Your Problem,

2
00:00:29,076 --> 00:00:31,196
Speaker 1: the show where I talk to people who are trying

3
00:00:31,236 --> 00:00:35,716
Speaker 1: to make technological progress. My guest today is Manola's. Kellis

4
00:00:36,196 --> 00:00:40,236
Speaker 1: Manola's is a professor of computer science at MIT, and

5
00:00:40,276 --> 00:00:44,956
Speaker 1: he works in computational biology. It's a field where researchers

6
00:00:44,996 --> 00:00:48,516
Speaker 1: take giant data sets relating to things like genetics and

7
00:00:48,636 --> 00:00:52,596
Speaker 1: health outcomes and try and understand basically what's going on,

8
00:00:52,916 --> 00:00:56,316
Speaker 1: things like what are the cellular mechanisms of disease and

9
00:00:56,476 --> 00:00:59,476
Speaker 1: how can we intervene to keep people healthy. In particular,

10
00:00:59,836 --> 00:01:04,836
Speaker 1: Minola's research focuses on genomics and a related field called epigenomics.

11
00:01:05,196 --> 00:01:07,236
Speaker 1: Here's how Manola's explains.

12
00:01:06,836 --> 00:01:13,276
Speaker 2: What that means. What's extraordinary with genomics is that we

13
00:01:13,356 --> 00:01:18,476
Speaker 2: can see beyond the limits of human imagination. We're talking

14
00:01:18,476 --> 00:01:21,916
Speaker 2: about millions of cells across hundreds of people, across thousands

15
00:01:21,916 --> 00:01:25,356
Speaker 2: of genes, and now we can now look at how

16
00:01:25,596 --> 00:01:29,916
Speaker 2: the single genome manifests in every cell type of the

17
00:01:29,996 --> 00:01:33,516
Speaker 2: human body in a slightly different way to create this

18
00:01:33,596 --> 00:01:38,836
Speaker 2: extraordinary symphony that is the human life, that is human thought,

19
00:01:38,876 --> 00:01:43,556
Speaker 2: that is human understanding, cognition, and every biological process that

20
00:01:44,236 --> 00:01:48,636
Speaker 2: ability to now start understanding the building blocks of how

21
00:01:48,996 --> 00:01:53,476
Speaker 2: this human genome manifests into all of these myriad of

22
00:01:53,516 --> 00:01:57,636
Speaker 2: cell types and their interactions and their combinations and their

23
00:01:57,676 --> 00:02:02,596
Speaker 2: coordination and their communication is what we can do for

24
00:02:02,636 --> 00:02:04,956
Speaker 2: the first time. They're also giving us the entry points

25
00:02:05,516 --> 00:02:10,116
Speaker 2: for understanding the basis of human variation, the basis of

26
00:02:10,156 --> 00:02:13,836
Speaker 2: human disease, and the basis for reversing human disease.

27
00:02:14,276 --> 00:02:17,116
Speaker 1: So that is the very big picture view of what

28
00:02:17,196 --> 00:02:20,436
Speaker 1: Manola's does. In our conversation, we got into a lot

29
00:02:20,476 --> 00:02:23,916
Speaker 1: more detail. For one thing, Manola's talked about his work

30
00:02:23,956 --> 00:02:28,196
Speaker 1: on obesity, and that work is based on epigenomics, which

31
00:02:28,276 --> 00:02:31,676
Speaker 1: is basically the way in which different genes are turned

32
00:02:31,836 --> 00:02:34,236
Speaker 1: on and off, and this turns out to be a

33
00:02:34,276 --> 00:02:37,396
Speaker 1: really big deal. Manola's and I also talked about his

34
00:02:37,556 --> 00:02:40,996
Speaker 1: work on Alzheimer's disease. In that part of the conversation,

35
00:02:41,356 --> 00:02:43,556
Speaker 1: he talked about how he and his colleagues are trying

36
00:02:43,596 --> 00:02:48,156
Speaker 1: to find these key biological pathways that contribute to lots

37
00:02:48,196 --> 00:02:50,676
Speaker 1: of different diseases, and how they're trying to come up

38
00:02:50,676 --> 00:02:55,036
Speaker 1: with drugs to target those pathways. We started our conversation

39
00:02:55,396 --> 00:02:58,516
Speaker 1: by talking about Manola's early work on the human genome,

40
00:02:58,676 --> 00:03:00,516
Speaker 1: which led to the work he's doing now.

41
00:03:00,636 --> 00:03:03,076
Speaker 2: So the human genome was mapped by K ninety nine

42
00:03:03,156 --> 00:03:05,396
Speaker 2: or two thousand and three, depending on how you count.

43
00:03:05,476 --> 00:03:08,196
Speaker 2: And then we had all of the nucleotides, all of

44
00:03:08,236 --> 00:03:13,676
Speaker 2: the letters through into billion letters. Then the hard part begins,

45
00:03:14,036 --> 00:03:15,796
Speaker 2: how do you make sense of that book? So that

46
00:03:15,876 --> 00:03:17,516
Speaker 2: was the Book of Life. So we had all of

47
00:03:17,516 --> 00:03:18,916
Speaker 2: the letters, how do you make sense of the book?

48
00:03:20,076 --> 00:03:26,596
Speaker 2: My own PhD was developing evolutionary signatures for understanding systematically

49
00:03:26,916 --> 00:03:30,676
Speaker 2: the human genome. So how do you recognize where are

50
00:03:30,716 --> 00:03:32,956
Speaker 2: the protein coding parts? What are the parts that code

51
00:03:32,956 --> 00:03:34,476
Speaker 2: for protein? We didn't even know.

52
00:03:34,676 --> 00:03:39,196
Speaker 1: And just to be clear, sort of non intuitively, most

53
00:03:39,236 --> 00:03:43,116
Speaker 1: of the human genome is not protein coding, right, Like

54
00:03:43,156 --> 00:03:46,676
Speaker 1: there's this very basic idea that like, oh, sure the genome,

55
00:03:46,676 --> 00:03:48,876
Speaker 1: that's what codes for proteins, but in fact, most of

56
00:03:48,916 --> 00:03:50,396
Speaker 1: the genome is not doing that.

57
00:03:51,196 --> 00:03:55,156
Speaker 2: Ninety eight percent of the human genome does not code

58
00:03:55,156 --> 00:03:55,676
Speaker 2: for protein.

59
00:03:55,796 --> 00:03:58,876
Speaker 1: It's wild. That is so nonintuitive, correct.

60
00:03:59,236 --> 00:04:02,916
Speaker 2: So in that mysterious ninety eight percent of the genome

61
00:04:03,396 --> 00:04:07,836
Speaker 2: lie control regents that are responsible for turning genes on

62
00:04:07,876 --> 00:04:13,756
Speaker 2: and off. And that's where ninety three percent of the

63
00:04:13,876 --> 00:04:16,556
Speaker 2: disease associated genetic variants are sitting.

64
00:04:17,156 --> 00:04:20,876
Speaker 1: Huh, it's not the genes that actually code for proteins,

65
00:04:20,956 --> 00:04:24,476
Speaker 1: it's the genes that control when are proteins made, when

66
00:04:24,476 --> 00:04:25,996
Speaker 1: are they not made, how much are they made.

67
00:04:26,076 --> 00:04:26,836
Speaker 2: That's exactly right.

68
00:04:26,916 --> 00:04:31,636
Speaker 1: Okay, so I get that in a broad sense. That's

69
00:04:31,676 --> 00:04:34,036
Speaker 1: sort of the state of affairs when you're coming into the.

70
00:04:34,036 --> 00:04:37,556
Speaker 2: Field's exactly right. So I wrote a series of papers,

71
00:04:37,636 --> 00:04:40,036
Speaker 2: both as a student and as a faculty member that

72
00:04:40,196 --> 00:04:44,276
Speaker 2: sought to then uncover how to even parse the genome,

73
00:04:44,316 --> 00:04:46,636
Speaker 2: how to even start understanding reading that book of life.

74
00:04:47,076 --> 00:04:49,796
Speaker 2: So that's one part. The second part is where the

75
00:04:49,836 --> 00:04:53,076
Speaker 2: regulatory motifs are. What are regulatory motifs. They are the

76
00:04:53,116 --> 00:04:57,556
Speaker 2: short words of the language of DNA that are bound

77
00:04:58,116 --> 00:05:02,196
Speaker 2: by regulators to turn genes on and off. So there's

78
00:05:02,236 --> 00:05:06,716
Speaker 2: these regulatory regions, and within these regions lie these words

79
00:05:07,676 --> 00:05:09,556
Speaker 2: which are the regulatory mode.

80
00:05:09,716 --> 00:05:13,756
Speaker 1: And just to be clear, the regulatory motifs are part

81
00:05:13,756 --> 00:05:17,396
Speaker 1: of what determine sort of when and how much different

82
00:05:17,436 --> 00:05:19,116
Speaker 1: genes express different proteins.

83
00:05:19,196 --> 00:05:21,636
Speaker 2: That's exactly right, that's exactly right. And that's where the

84
00:05:21,716 --> 00:05:24,476
Speaker 2: human epigenome comes in. So what we needed to now

85
00:05:24,556 --> 00:05:28,156
Speaker 2: understand is how that genome turns to life. So you

86
00:05:28,196 --> 00:05:30,756
Speaker 2: can think of the epigenome as the living genome, as

87
00:05:30,796 --> 00:05:33,836
Speaker 2: the genome. There's the genome itself is static. It's just

88
00:05:33,876 --> 00:05:36,316
Speaker 2: the book the tablets, if you wish that Moses brought

89
00:05:36,356 --> 00:05:40,076
Speaker 2: down from the mountain, and then the epigenome is the

90
00:05:40,276 --> 00:05:43,796
Speaker 2: music that gets played from the orchestra. The epigenome tells

91
00:05:43,836 --> 00:05:47,116
Speaker 2: you which parts are active in the brain and the liver,

92
00:05:47,236 --> 00:05:49,236
Speaker 2: and the heart and the muscle and so and so forth.

93
00:05:49,796 --> 00:05:53,156
Speaker 1: So your work on the epigenome is really interesting to me.

94
00:05:53,676 --> 00:05:56,356
Speaker 1: And I know you've done some work on obesity, and

95
00:05:56,396 --> 00:05:58,996
Speaker 1: the epigenome tell me about that.

96
00:05:59,436 --> 00:06:02,796
Speaker 2: The strongest genetic association with obesity sits in one gene

97
00:06:03,476 --> 00:06:09,676
Speaker 2: called FTO, and FTO was renamed fat and obesity associated

98
00:06:09,716 --> 00:06:13,716
Speaker 2: after that discovery, and it remained mysterious for seven years.

99
00:06:14,036 --> 00:06:16,436
Speaker 2: People had no idea how that gene works.

100
00:06:16,836 --> 00:06:17,876
Speaker 1: You just saw correlate.

101
00:06:17,956 --> 00:06:18,676
Speaker 2: There was a correlation.

102
00:06:18,836 --> 00:06:19,516
Speaker 1: There was a correlation.

103
00:06:19,836 --> 00:06:22,316
Speaker 2: Just the problem of genetics and the beauty of genetics.

104
00:06:22,356 --> 00:06:24,716
Speaker 2: The beauty of genetics is that it tells you what

105
00:06:24,996 --> 00:06:28,516
Speaker 2: region is responsible for disease. Regardless of how it functions.

106
00:06:29,156 --> 00:06:32,036
Speaker 2: The downside is that it after he tells.

107
00:06:31,756 --> 00:06:32,836
Speaker 1: You it's the same thing.

108
00:06:33,596 --> 00:06:36,076
Speaker 2: After it tells you that he has a role, you

109
00:06:36,116 --> 00:06:40,596
Speaker 2: have no idea how it functions. And what we showed

110
00:06:41,196 --> 00:06:45,956
Speaker 2: in our work is that that region doesn't affect the

111
00:06:46,076 --> 00:06:47,276
Speaker 2: FTO gene at all.

112
00:06:47,356 --> 00:06:50,556
Speaker 1: So like in the middle of a gene, there is

113
00:06:50,596 --> 00:06:56,316
Speaker 1: this whatever series of nucleotides, but those those nucleotides are

114
00:06:56,356 --> 00:06:58,396
Speaker 1: just randomly in the middle of that gene and actually

115
00:06:58,436 --> 00:07:00,156
Speaker 1: have nothing to do with that gene. I didn't even

116
00:07:00,156 --> 00:07:01,276
Speaker 1: know you could do that.

117
00:07:01,236 --> 00:07:07,156
Speaker 2: Fairly, you can't. So there are eighty nine differences, eighty

118
00:07:07,236 --> 00:07:11,876
Speaker 2: nine common variants, common genetic variants that are all coinherited.

119
00:07:12,236 --> 00:07:14,596
Speaker 2: If you get a here, you get all of the

120
00:07:14,636 --> 00:07:19,116
Speaker 2: other you know, actage, you get that passage. If you

121
00:07:19,116 --> 00:07:22,116
Speaker 2: get that package, it spans fifty thousand letters. But there

122
00:07:22,156 --> 00:07:25,956
Speaker 2: are only eighty nine differences in these fifty thousand letters. Wow,

123
00:07:26,396 --> 00:07:32,116
Speaker 2: and these will increase your body weight by one standard deviation,

124
00:07:33,676 --> 00:07:35,676
Speaker 2: which is like how much it's like nine pounds, Like

125
00:07:35,756 --> 00:07:38,636
Speaker 2: it's a lot, okay. So so basically what that does

126
00:07:39,196 --> 00:07:42,756
Speaker 2: is that it functions somehow to increase your risk for

127
00:07:42,756 --> 00:07:46,516
Speaker 2: a basits, it's like the strongest genetic association before. And

128
00:07:47,276 --> 00:07:50,356
Speaker 2: what we reason is, how could it be acting. It

129
00:07:50,396 --> 00:07:52,596
Speaker 2: could be acting in your brain to decide whether you

130
00:07:52,716 --> 00:07:55,116
Speaker 2: like sweets or salting. It could be acting your muscle

131
00:07:55,156 --> 00:07:57,196
Speaker 2: to make you more fit or less fit. It could

132
00:07:57,236 --> 00:08:00,916
Speaker 2: be asking in your digestives. So we basically said, okay,

133
00:08:00,996 --> 00:08:03,716
Speaker 2: well that's speculation. Let's look at the data. And we

134
00:08:03,796 --> 00:08:05,716
Speaker 2: looked at the data and we found that there was

135
00:08:05,796 --> 00:08:11,316
Speaker 2: this massive control region that was active in mesenchymal stem

136
00:08:11,356 --> 00:08:13,916
Speaker 2: cells what are mesimo cells and sells. They are the

137
00:08:13,956 --> 00:08:21,036
Speaker 2: progenitors of brown fat and white fat. Now, white fat

138
00:08:21,396 --> 00:08:24,836
Speaker 2: is white because it's full of lipids. That's where all

139
00:08:24,876 --> 00:08:28,276
Speaker 2: the calories are stored. Brown fat is brown because of

140
00:08:28,316 --> 00:08:31,076
Speaker 2: all of the iron in the mitochondria. That's where the

141
00:08:31,076 --> 00:08:34,156
Speaker 2: calories are burned. So it turns out that our fat

142
00:08:34,196 --> 00:08:38,076
Speaker 2: cells make a developmental decision in their first three days

143
00:08:38,076 --> 00:08:41,756
Speaker 2: of differentiation to go down the white path lineage or

144
00:08:41,916 --> 00:08:45,396
Speaker 2: the brown path lineage. And what the white fat does

145
00:08:45,636 --> 00:08:51,796
Speaker 2: is it stores energies and brown burns energies. So it

146
00:08:51,836 --> 00:08:55,156
Speaker 2: turns out that I'm actually homozygous risk for the store

147
00:08:55,196 --> 00:08:58,796
Speaker 2: calories position, which is the obesity risk.

148
00:08:58,996 --> 00:09:00,476
Speaker 1: So you have the obesity.

149
00:09:00,596 --> 00:09:03,036
Speaker 2: I have two copies of the obesity risk. My wife

150
00:09:03,076 --> 00:09:06,276
Speaker 2: has zero. I can tell you, you know, we look

151
00:09:06,356 --> 00:09:12,196
Speaker 2: very different. Fair So we basically realize that it sits

152
00:09:12,196 --> 00:09:16,036
Speaker 2: in the progenitor cells of white and brown flat and

153
00:09:16,116 --> 00:09:19,836
Speaker 2: then we could show that the true target was not

154
00:09:19,956 --> 00:09:23,676
Speaker 2: the ftogene at all. It was instead two other genes

155
00:09:23,716 --> 00:09:28,116
Speaker 2: that are sitting one point two million letters away from

156
00:09:28,116 --> 00:09:31,796
Speaker 2: this region and six hundred thousand letters away, and those

157
00:09:31,836 --> 00:09:37,756
Speaker 2: genes turned out to be master controllers of thermogenesis. They

158
00:09:37,796 --> 00:09:44,116
Speaker 2: are basically switching your metabolic state. So my cells are

159
00:09:44,236 --> 00:09:48,516
Speaker 2: stuck on the store position and my wife cells are

160
00:09:48,556 --> 00:09:49,716
Speaker 2: stuck on the burn position.

161
00:09:50,596 --> 00:09:53,676
Speaker 1: And so what is the relationship between the genes that

162
00:09:53,716 --> 00:09:59,116
Speaker 1: are acting here and this this you know, package variant

163
00:09:59,116 --> 00:10:00,636
Speaker 1: that is far away from them.

164
00:10:00,756 --> 00:10:05,196
Speaker 2: It comes back to the epigena. So our DNA is

165
00:10:05,316 --> 00:10:10,596
Speaker 2: stored inside a tiny little space. The way that gene

166
00:10:10,596 --> 00:10:13,476
Speaker 2: regulation works is that you have these control regions that

167
00:10:13,516 --> 00:10:17,716
Speaker 2: are scattered throughout the region of every gene that are

168
00:10:17,796 --> 00:10:21,076
Speaker 2: linked together to that gene in three dimensions. So they

169
00:10:21,156 --> 00:10:22,836
Speaker 2: do around and.

170
00:10:22,756 --> 00:10:24,796
Speaker 1: So it's it's far away. If you think of it

171
00:10:24,836 --> 00:10:27,636
Speaker 1: as a strand but in three dimensional space, right there,

172
00:10:27,996 --> 00:10:32,276
Speaker 1: three dimension pats right, Ah, that's satisfying.

173
00:10:32,396 --> 00:10:36,036
Speaker 2: And when we took these genes and we modulated them,

174
00:10:36,636 --> 00:10:41,436
Speaker 2: we show that you can turn off one gene in mouse,

175
00:10:42,276 --> 00:10:46,956
Speaker 2: in specifically the adipocytes of mouse with a dominant negative

176
00:10:46,996 --> 00:10:51,316
Speaker 2: cus of fat cells with a dominant negative construct, and

177
00:10:51,476 --> 00:10:56,236
Speaker 2: that turned the mouse fifty percent leaner. They eat the

178
00:10:56,276 --> 00:10:59,556
Speaker 2: same amount, they exercise the same amount, but they burn

179
00:10:59,636 --> 00:11:03,756
Speaker 2: calories when they're awake and they burn calories when they're sleeping.

180
00:11:05,036 --> 00:11:08,116
Speaker 2: And what's really fascinated with that story is that the

181
00:11:08,276 --> 00:11:13,116
Speaker 2: variant associated with obesity is at two percent frequency in Africa,

182
00:11:13,996 --> 00:11:17,716
Speaker 2: but forty two percent frequency in Europe and forty four

183
00:11:17,756 --> 00:11:22,276
Speaker 2: percent frequency in Southeast Asia. So it rose from two

184
00:11:22,316 --> 00:11:27,196
Speaker 2: percent to forty four percent maybe because of positive selection.

185
00:11:27,996 --> 00:11:31,076
Speaker 2: Maybe it was beneficial to be able to store every

186
00:11:31,116 --> 00:11:31,676
Speaker 2: kind of.

187
00:11:31,556 --> 00:11:34,276
Speaker 1: Places where food is, where you have food is scarce

188
00:11:34,316 --> 00:11:36,436
Speaker 1: in moments of famine, exactly.

189
00:11:36,036 --> 00:11:38,276
Speaker 2: In the out of Africa event, this may have been

190
00:11:38,276 --> 00:11:40,916
Speaker 2: selected for. Or in the you know, ice ages, it

191
00:11:40,956 --> 00:11:44,316
Speaker 2: may have been selected for. And it's only after World

192
00:11:44,316 --> 00:11:49,036
Speaker 2: War two that this variant became associated with obesity.

193
00:11:48,636 --> 00:11:51,036
Speaker 1: Because food became so abundant.

194
00:11:50,556 --> 00:11:54,436
Speaker 2: And we stopped exercising as much. So it's fascinating to

195
00:11:54,436 --> 00:11:58,196
Speaker 2: see how the environmental shift led to a new genetic

196
00:11:58,196 --> 00:12:02,956
Speaker 2: association which is now plaguing our society, and of course

197
00:12:02,996 --> 00:12:07,316
Speaker 2: the hope that by understanding the circuit systematically, we can

198
00:12:07,396 --> 00:12:13,676
Speaker 2: now solve so many different circuits and ultimately so many

199
00:12:13,676 --> 00:12:17,076
Speaker 2: different pathways and ultimately so many different disorders.

200
00:12:19,636 --> 00:12:22,956
Speaker 1: In a minute, Manola's describes how he and his colleagues

201
00:12:22,996 --> 00:12:26,476
Speaker 1: are trying to turn their genomic research into new medicines.

202
00:12:35,476 --> 00:12:36,436
Speaker 1: That's the end of the ads.

203
00:12:36,876 --> 00:12:38,036
Speaker 2: Now we're going back to the show.

204
00:12:39,036 --> 00:12:41,396
Speaker 1: Another area where Manola's and his colleagues have done a

205
00:12:41,436 --> 00:12:44,796
Speaker 1: lot of work is on Alzheimer's disease. They looked at

206
00:12:44,836 --> 00:12:48,876
Speaker 1: a common genetic variant called apo E four. People with

207
00:12:48,956 --> 00:12:51,556
Speaker 1: two copies of this variant have a much much higher

208
00:12:51,636 --> 00:12:54,956
Speaker 1: risk of getting Alzheimer's, and Manola's and his colleagues were

209
00:12:54,956 --> 00:12:57,996
Speaker 1: trying to figure out why. They found that having this

210
00:12:58,116 --> 00:13:02,436
Speaker 1: Apoe four variant was linked to problems with moving cholesterol

211
00:13:02,636 --> 00:13:07,956
Speaker 1: around in the brain, a process called cholesterol transport. Then

212
00:13:08,156 --> 00:13:11,356
Speaker 1: they did experiments and mice that found that drugs that

213
00:13:11,436 --> 00:13:17,036
Speaker 1: restore cholesterol transport actually restored cognition in the mice. Now

214
00:13:17,156 --> 00:13:21,316
Speaker 1: that's in mice, and Alzheimer's is a notoriously difficult disease

215
00:13:21,396 --> 00:13:25,316
Speaker 1: to treat in humans. So I asked Minolas what it

216
00:13:25,356 --> 00:13:28,116
Speaker 1: will take to move his research from mice to humans,

217
00:13:28,476 --> 00:13:31,956
Speaker 1: and his answer was really interesting. It pointed not only

218
00:13:31,956 --> 00:13:35,276
Speaker 1: two ideas about treating Alzheimer's, but to bigger ideas about

219
00:13:35,316 --> 00:13:37,116
Speaker 1: treating human disease more generally.

220
00:13:38,236 --> 00:13:39,916
Speaker 2: The way that I'm thinking about this, the way that

221
00:13:39,956 --> 00:13:43,116
Speaker 2: our team is thinking about these, is how do we

222
00:13:43,276 --> 00:13:49,476
Speaker 2: enable personalized medicine and precision medicine. Namely, Alzheimer's is not

223
00:13:49,476 --> 00:13:51,956
Speaker 2: going to be only about transport. It's going to be

224
00:13:51,996 --> 00:13:56,276
Speaker 2: a combination. Every person has some combination of these regulations.

225
00:13:56,716 --> 00:14:00,596
Speaker 2: A point four is the strongest genetic risk, but there

226
00:14:00,596 --> 00:14:03,676
Speaker 2: are many others. And the question is how do we

227
00:14:03,996 --> 00:14:07,476
Speaker 2: now systematically take a person with Alzheimer's, or take a

228
00:14:07,516 --> 00:14:12,036
Speaker 2: family with risk, develop treatments that are either directly addressing

229
00:14:12,036 --> 00:14:17,196
Speaker 2: the root causes rather than treating the symptoms, and are

230
00:14:17,356 --> 00:14:22,556
Speaker 2: not only preventative but adapted to every family and every person.

231
00:14:22,876 --> 00:14:25,116
Speaker 1: And just to be clear, like having you know, two

232
00:14:25,196 --> 00:14:30,196
Speaker 1: copies of the APO four lil is neither necessary nor

233
00:14:30,196 --> 00:14:32,796
Speaker 1: sufficient to get Alzheimer's. Right, that's exactly both of them

234
00:14:32,796 --> 00:14:34,436
Speaker 1: and not get it. You can have neither of them

235
00:14:34,476 --> 00:14:37,756
Speaker 1: and get it. So it's exactly so complicated hard.

236
00:14:37,836 --> 00:14:41,676
Speaker 2: So, as with everything with human disease, genetics is not destiny.

237
00:14:42,076 --> 00:14:45,836
Speaker 2: Genetics is a predisposition, and there are environmental factors. There

238
00:14:45,876 --> 00:14:50,316
Speaker 2: are behavioral factors, there are nutritional exercise factors, there are

239
00:14:50,356 --> 00:14:52,836
Speaker 2: socio economic factors. There's so many other factors that are

240
00:14:52,876 --> 00:14:58,836
Speaker 2: affecting how your genetics will manifest ultimately into disease. But

241
00:14:59,076 --> 00:15:02,316
Speaker 2: now the question is for every person, how do we

242
00:15:02,676 --> 00:15:06,636
Speaker 2: create a drug? And it's not going to be feasible

243
00:15:06,756 --> 00:15:11,276
Speaker 2: economically or in any other way to create one pill

244
00:15:11,316 --> 00:15:13,716
Speaker 2: for each person. The way that we're going to enable

245
00:15:13,716 --> 00:15:17,916
Speaker 2: personalized medicine is by understanding what are the hallmarks of disease,

246
00:15:18,316 --> 00:15:20,916
Speaker 2: what are the hallmarks of Alzheimer's, the wholemarks of obesity,

247
00:15:20,996 --> 00:15:23,916
Speaker 2: the whole moods of diabetes, the hallmarks of cardiac disorders,

248
00:15:24,276 --> 00:15:28,436
Speaker 2: and develop therapeutics for every one of those hallmarks. So

249
00:15:28,516 --> 00:15:31,836
Speaker 2: think of it as an arsenal of twelve or twenty

250
00:15:32,076 --> 00:15:35,076
Speaker 2: different drugs for Alzheimer's that you're going to be taking

251
00:15:35,076 --> 00:15:37,156
Speaker 2: a combination of it.

252
00:15:37,196 --> 00:15:41,156
Speaker 1: Seems like oncology is already some way down that road, right,

253
00:15:41,196 --> 00:15:44,876
Speaker 1: I mean, you know her two positive breast cancers have

254
00:15:44,996 --> 00:15:47,036
Speaker 1: certain drugs that target them that sort of thing, right,

255
00:15:47,116 --> 00:15:47,836
Speaker 1: is that the model?

256
00:15:48,596 --> 00:15:51,876
Speaker 2: That's exactly the model. So the hallmarks of cancer have

257
00:15:52,036 --> 00:15:55,116
Speaker 2: been the way of thinking about cancer for twenty plus

258
00:15:55,236 --> 00:15:58,716
Speaker 2: years now. And the difference in cancer is the following.

259
00:15:59,476 --> 00:16:04,356
Speaker 2: Cancer is subject to positive selection. What does that mean?

260
00:16:04,716 --> 00:16:09,236
Speaker 2: That means that because it's a replicative disorder where the cell,

261
00:16:09,356 --> 00:16:12,836
Speaker 2: the cancer cells make more of themselves. If a cell

262
00:16:12,916 --> 00:16:17,636
Speaker 2: acquires a mutation that allows it to replicate faster, you

263
00:16:17,716 --> 00:16:21,396
Speaker 2: will have more of that cell. So you are subject

264
00:16:21,476 --> 00:16:25,716
Speaker 2: to positive selection where the bad mutations are increasing in

265
00:16:25,756 --> 00:16:32,556
Speaker 2: frequency in every generation of the cancer. By contrast, most

266
00:16:32,596 --> 00:16:37,556
Speaker 2: complex disorders are subject to purifying selection, where the mutations

267
00:16:37,596 --> 00:16:40,916
Speaker 2: that are responsible for them are maintained at low frequency

268
00:16:40,916 --> 00:16:41,516
Speaker 2: by evolution.

269
00:16:42,556 --> 00:16:42,836
Speaker 1: Huh.

270
00:16:43,356 --> 00:16:47,556
Speaker 2: So it's working at the opposite ends of the evolutionary spectrum.

271
00:16:47,836 --> 00:16:50,796
Speaker 2: So cancer has a small number of genes that drive

272
00:16:50,956 --> 00:16:55,236
Speaker 2: the disorder. Complex traits have thousands of genes that are

273
00:16:55,356 --> 00:16:59,396
Speaker 2: maintained at low frequency or at weak effects.

274
00:16:59,836 --> 00:17:03,276
Speaker 1: Except that sounds much harder. It's harder to figure out

275
00:17:03,316 --> 00:17:04,116
Speaker 1: what's going on harder.

276
00:17:05,236 --> 00:17:08,076
Speaker 2: But the saving grace is that even though you have

277
00:17:08,196 --> 00:17:12,356
Speaker 2: extreme heterogeneity in the number of drivers, for every one

278
00:17:12,356 --> 00:17:19,476
Speaker 2: of these disorders, they coalesce, they cluster, they converge in

279
00:17:19,556 --> 00:17:24,276
Speaker 2: a small number of recurrent pathways, and these are the hallmarks.

280
00:17:24,676 --> 00:17:24,876
Speaker 1: Huh.

281
00:17:25,396 --> 00:17:28,716
Speaker 2: So you can find multiple genes associated with lipid transport,

282
00:17:28,996 --> 00:17:32,036
Speaker 2: you can find multiple genes associated with new inflammation with

283
00:17:32,156 --> 00:17:33,156
Speaker 2: DNA damage, so.

284
00:17:33,116 --> 00:17:35,876
Speaker 1: You target the sort of pathways where they converge.

285
00:17:35,956 --> 00:17:37,876
Speaker 2: That's exactly right. So we're not going to make a

286
00:17:38,036 --> 00:17:40,916
Speaker 2: drug for Alzheimer's that we might make a drug for

287
00:17:41,036 --> 00:17:44,716
Speaker 2: DNA damage, a drug for lipid metabolism, a drug for

288
00:17:44,796 --> 00:17:48,076
Speaker 2: cholesterol transport, et cetera. And that's what we're working.

289
00:17:48,356 --> 00:17:50,916
Speaker 1: That's satisfying. That's a satisfying explanation.

290
00:17:51,316 --> 00:17:54,876
Speaker 2: It basically says that it is a limited number. There's

291
00:17:54,916 --> 00:17:57,076
Speaker 2: a billion people in the planet. We're not going to

292
00:17:57,116 --> 00:17:59,356
Speaker 2: have a billion drugs. What we're going to have it's

293
00:17:59,396 --> 00:18:02,796
Speaker 2: a small number of drugs, one for each pathway, and

294
00:18:02,916 --> 00:18:07,276
Speaker 2: these are sometimes going to be actually reused between different disorders.

295
00:18:07,836 --> 00:18:10,796
Speaker 2: So we work on cardie disorders, we're finding the same

296
00:18:10,916 --> 00:18:16,196
Speaker 2: genes underlying Alzheimer's, and specifically the lipid and cholesterol component

297
00:18:16,676 --> 00:18:20,196
Speaker 2: are in fact reused in the heart disease. And again

298
00:18:20,236 --> 00:18:25,356
Speaker 2: it's about lipids. It's about saturation of the fat stores

299
00:18:25,396 --> 00:18:28,036
Speaker 2: of an individual and now the lipid escaping into the

300
00:18:28,036 --> 00:18:32,036
Speaker 2: blacks into the bloodstream, forming these plaques that will then

301
00:18:32,196 --> 00:18:35,756
Speaker 2: cause heart you know, failure and heart damage and so

302
00:18:35,796 --> 00:18:38,836
Speaker 2: and so forth. So that's where we're at.

303
00:18:39,036 --> 00:18:43,236
Speaker 1: So is there. I mean, the dream is that there

304
00:18:43,316 --> 00:18:46,476
Speaker 1: is some dysfunction that is common to all these different

305
00:18:46,516 --> 00:18:51,156
Speaker 1: diseases that you could target, right, Like, I mean, the

306
00:18:51,276 --> 00:18:53,956
Speaker 1: naive dream is find the cure for everything, or not everything,

307
00:18:53,996 --> 00:18:55,596
Speaker 1: but find the cure for a lot of things, or

308
00:18:55,636 --> 00:18:59,316
Speaker 1: at least find a drug that will reduce risks of

309
00:18:59,396 --> 00:19:02,916
Speaker 1: many different bad things, right, I mean, is that plausible

310
00:19:02,996 --> 00:19:05,476
Speaker 1: or am I just naive in going there? From what

311
00:19:05,516 --> 00:19:05,996
Speaker 1: you're saying.

312
00:19:06,276 --> 00:19:11,796
Speaker 2: So you're right that some of the time these pathways

313
00:19:11,796 --> 00:19:14,796
Speaker 2: that we're finding are going to be helping in multiple frauds,

314
00:19:15,636 --> 00:19:18,476
Speaker 2: And then that's absolutely the dream. We should basically start

315
00:19:18,596 --> 00:19:21,036
Speaker 2: not with what is the worst disease, but maybe what

316
00:19:21,116 --> 00:19:23,596
Speaker 2: is the best pathway that if we fix that one,

317
00:19:23,676 --> 00:19:26,156
Speaker 2: we're going to have an impact on most diseases.

318
00:19:25,956 --> 00:19:28,636
Speaker 1: Right, like the highest return on investments for example.

319
00:19:28,676 --> 00:19:30,556
Speaker 2: Like, Yeah, that's a great way to think about it.

320
00:19:31,956 --> 00:19:35,276
Speaker 2: But the way that I would say is that for

321
00:19:35,636 --> 00:19:38,236
Speaker 2: each person, this might be a different molecule.

322
00:19:39,876 --> 00:19:42,396
Speaker 1: So now I'm not hopeful.

323
00:19:43,596 --> 00:19:47,036
Speaker 2: But that with a small number of these molecules, say

324
00:19:47,076 --> 00:19:48,956
Speaker 2: one hundred, one hundred and fifty two hundred molecules.

325
00:19:48,956 --> 00:19:50,436
Speaker 1: When you say molecule, you mean drug.

326
00:19:50,356 --> 00:19:52,836
Speaker 2: I mean trust, might I mean drust. Yeah, Basically that

327
00:19:52,876 --> 00:19:54,956
Speaker 2: there's going to be a small number of pathways and

328
00:19:55,036 --> 00:19:59,396
Speaker 2: a small number of these modulators, and that those are

329
00:19:59,396 --> 00:20:01,876
Speaker 2: going to be mixed and matched in each person to

330
00:20:02,116 --> 00:20:04,996
Speaker 2: then target a communatorially large number of people.

331
00:20:05,076 --> 00:20:07,556
Speaker 1: Yeah, it just got hard. I know, I know biology

332
00:20:07,676 --> 00:20:10,196
Speaker 1: is hard, but I got up to for a second.

333
00:20:10,836 --> 00:20:12,596
Speaker 2: There's not going to be a single silver bullet for

334
00:20:13,076 --> 00:20:15,276
Speaker 2: all of those. In fact, for any one of these diseases,

335
00:20:15,276 --> 00:20:17,996
Speaker 2: there's no silver bullet. But the moment you build your

336
00:20:18,076 --> 00:20:20,956
Speaker 2: panelbly of fifty silver bullets, then you're going to be

337
00:20:20,996 --> 00:20:24,196
Speaker 2: hitting two hundred diseases. That's the beauty of it.

338
00:20:24,396 --> 00:20:26,996
Speaker 1: Fifty bronze bo there's no silver bullet, but maybe.

339
00:20:26,836 --> 00:20:28,236
Speaker 2: You can find it for hearts exactly right.

340
00:20:30,356 --> 00:20:43,916
Speaker 1: We'll be back in a minute with the lightning round. Now,

341
00:20:43,956 --> 00:20:46,516
Speaker 1: let's get back to the show. I read that you

342
00:20:46,676 --> 00:20:48,876
Speaker 1: have been an author on more than two hundred and

343
00:20:48,956 --> 00:20:52,156
Speaker 1: thirty papers, which is a lot. Which one was the

344
00:20:52,156 --> 00:20:52,636
Speaker 1: most fun?

345
00:20:52,756 --> 00:20:54,196
Speaker 2: Oh? You know what, don't I tell you about my

346
00:20:54,316 --> 00:20:54,876
Speaker 2: very first one?

347
00:20:54,916 --> 00:20:55,196
Speaker 1: Sure?

348
00:20:56,756 --> 00:20:59,876
Speaker 2: And the very first paper was published in c graph

349
00:21:00,036 --> 00:21:02,196
Speaker 2: and it now has like two thousand citations, And it

350
00:21:02,236 --> 00:21:05,916
Speaker 2: was about how do we reconstruct the surface of an

351
00:21:05,956 --> 00:21:09,836
Speaker 2: object from a cloud of points? So you can basically

352
00:21:09,916 --> 00:21:12,476
Speaker 2: use laser scanning to sort of figure out points in

353
00:21:12,516 --> 00:21:14,636
Speaker 2: three D and then the question is what is the

354
00:21:14,676 --> 00:21:17,516
Speaker 2: surface that goes between them. I've always been fascinated with

355
00:21:17,556 --> 00:21:19,716
Speaker 2: three D space, so it was very fun for me

356
00:21:19,796 --> 00:21:22,036
Speaker 2: to just like you know, as a kid, basically as

357
00:21:22,116 --> 00:21:25,636
Speaker 2: as a freshman at to work on such a project

358
00:21:25,716 --> 00:21:28,916
Speaker 2: and then showing up at the conference. He was in Disneyland,

359
00:21:29,196 --> 00:21:30,876
Speaker 2: so it was my first time in Disneyland as an

360
00:21:30,916 --> 00:21:31,996
Speaker 2: author of a vapor.

361
00:21:31,836 --> 00:21:35,156
Speaker 1: Sounds relevant for motion capture, not knowing anything about it.

362
00:21:35,196 --> 00:21:38,636
Speaker 1: When I think of, like, you know, people, the way

363
00:21:38,636 --> 00:21:40,636
Speaker 1: they make movies now exactly as they put a bunch

364
00:21:40,676 --> 00:21:42,876
Speaker 1: of censors on people and they move around and then

365
00:21:42,916 --> 00:21:45,276
Speaker 1: you can turn them into a dragon or whatever you want.

366
00:21:45,316 --> 00:21:48,636
Speaker 2: Yeah, that's exactly right. So you know that paper has

367
00:21:48,676 --> 00:21:50,756
Speaker 2: been quite influential and used for a lot of a

368
00:21:50,796 --> 00:21:51,676
Speaker 2: lot of different things.

369
00:21:52,076 --> 00:21:54,276
Speaker 1: What's the most overrated Greek island?

370
00:21:54,316 --> 00:21:55,756
Speaker 2: Oh my god, I can tell you about the most

371
00:21:55,836 --> 00:21:59,676
Speaker 2: underrated Santorini. Definitely not overrated tons of people, but worth

372
00:21:59,836 --> 00:22:02,276
Speaker 2: every time. I can tell you about my first day

373
00:22:02,316 --> 00:22:05,516
Speaker 2: in Santorini, which is I walked out on this balcony

374
00:22:05,796 --> 00:22:07,716
Speaker 2: and I asked the owner of the restaurant if I

375
00:22:07,756 --> 00:22:09,036
Speaker 2: can take a look at the view and I'm not

376
00:22:09,436 --> 00:22:12,996
Speaker 2: order anything. He said, please be my guest, and I

377
00:22:13,036 --> 00:22:15,196
Speaker 2: walked out, and ten minutes later, I'm like, I can't leave.

378
00:22:15,236 --> 00:22:18,556
Speaker 2: I'm gonna have to order. He tells me, ten years ago,

379
00:22:18,596 --> 00:22:19,836
Speaker 2: I came here to look at the view.

380
00:22:19,916 --> 00:22:21,756
Speaker 1: I want you to throw a little bit of shade.

381
00:22:21,836 --> 00:22:23,556
Speaker 1: I want you to get in a little bit of drug.

382
00:22:23,596 --> 00:22:23,996
Speaker 2: Can't.

383
00:22:24,116 --> 00:22:26,636
Speaker 1: What's one place in Greece I should not cannot.

384
00:22:28,356 --> 00:22:32,676
Speaker 2: It's not possible. I mean, you know, if you keep insisting,

385
00:22:32,676 --> 00:22:34,836
Speaker 2: I'll give you another twenty amazing places to visit.

386
00:22:35,236 --> 00:22:38,116
Speaker 1: Well, that's fair, that's fair. I did what I could do.

387
00:22:38,596 --> 00:22:41,396
Speaker 1: If everything goes well, what problem will you be trying

388
00:22:41,396 --> 00:22:43,036
Speaker 1: to solve in five years?

389
00:22:43,676 --> 00:22:46,836
Speaker 2: I think what I'm trying to solve now of actually

390
00:22:47,756 --> 00:22:53,156
Speaker 2: creating these drugs in such a modular, AI driven, personalized,

391
00:22:53,796 --> 00:22:58,076
Speaker 2: reusable way, centered on pathways. That's going to keep me

392
00:22:58,116 --> 00:23:01,076
Speaker 2: busy for a long time. And I hope that in

393
00:23:01,116 --> 00:23:05,876
Speaker 2: five years we have actually sold a big chunk of

394
00:23:05,916 --> 00:23:10,036
Speaker 2: the platform and that we have a few drugs in

395
00:23:10,076 --> 00:23:13,076
Speaker 2: clinical trials. So you know, my dream needs to take

396
00:23:13,116 --> 00:23:16,196
Speaker 2: all of these circuits that we have uncovered and make

397
00:23:16,236 --> 00:23:18,316
Speaker 2: a difference for humanity, make a difference for you know,

398
00:23:18,356 --> 00:23:21,116
Speaker 2: my fellow beings. That's my big goal.

399
00:23:21,596 --> 00:23:23,596
Speaker 1: Great, it's fun to talk to you.

400
00:23:23,996 --> 00:23:26,516
Speaker 2: I learned a lot, such a pleasure, thank you, and

401
00:23:26,556 --> 00:23:29,516
Speaker 2: I love that you're fearless. You're like, well, we're gonna

402
00:23:29,596 --> 00:23:32,596
Speaker 2: jump into this new topic and find it all about it.

403
00:23:36,356 --> 00:23:39,676
Speaker 1: Man nola's Kellis is a professor of computer science at MIT.

404
00:23:40,716 --> 00:23:45,036
Speaker 1: Today's show was produced by Edith Russelo, edited by Karen Chakerji,

405
00:23:45,396 --> 00:23:48,796
Speaker 1: and engineered by Sarah Bruguer. You can email us at

406
00:23:48,916 --> 00:23:53,196
Speaker 1: problem at pushkin dot FM. I'm Jacob Goldstein. One last

407
00:23:53,236 --> 00:23:55,316
Speaker 1: thing we are going to be taking a break for

408
00:23:55,396 --> 00:23:57,436
Speaker 1: a couple of weeks, but we'll be back with new

409
00:23:57,476 --> 00:24:01,436
Speaker 1: shows in early twenty twenty four. Thanks for listening, Happy

410
00:24:01,436 --> 00:24:10,436
Speaker 1: New Year, that t