TEDSummit 2019
Sara-Jane Dunn: The next software revolution: programming biological cells
Filmed:
Readability: 5.2
1,706,478 views
The cells in your body are like computer software: they're "programmed" to carry out specific functions at specific times. If we can better understand this process, we could unlock the ability to reprogram cells ourselves, says computational biologist Sara-Jane Dunn. In a talk from the cutting-edge of science, she explains how her team is studying embryonic stem cells to gain a new understanding of the biological programs that power life -- and develop "living software" that could transform medicine, agriculture and energy.
Sara-Jane Dunn - Computational biologist
Sara-Jane Dunn is a scientist working at the interface between biology and computation, using mathematics and computational analysis to make sense of how living systems process information. Full bio
Sara-Jane Dunn is a scientist working at the interface between biology and computation, using mathematics and computational analysis to make sense of how living systems process information. Full bio
Double-click the English transcript below to play the video.
00:12
The second half of the last century
was completely defined
was completely defined
0
750
4509
00:17
by a technological revolution:
1
5283
1999
00:19
the software revolution.
2
7306
1435
00:21
The ability to program electrons
on a material called silicon
on a material called silicon
3
9313
4808
00:26
made possible technologies,
companies and industries
companies and industries
4
14145
3073
00:29
that were at one point
unimaginable to many of us,
unimaginable to many of us,
5
17242
3977
00:33
but which have now fundamentally changed
the way the world works.
the way the world works.
6
21243
3915
00:38
The first half of this century, though,
7
26158
1921
00:40
is going to be transformed
by a new software revolution:
by a new software revolution:
8
28103
3978
00:44
the living software revolution.
9
32105
2435
00:46
And this will be powered by the ability
to program biochemistry
to program biochemistry
10
34921
4050
00:50
on a material called biology.
11
38995
2295
00:53
And doing so will enable us to harness
the properties of biology
the properties of biology
12
41314
4141
00:57
to generate new kinds of therapies,
13
45479
2656
01:00
to repair damaged tissue,
14
48159
1868
01:02
to reprogram faulty cells
15
50051
2725
01:04
or even build programmable
operating systems out of biochemistry.
operating systems out of biochemistry.
16
52800
4554
01:10
If we can realize this --
and we do need to realize it --
and we do need to realize it --
17
58420
3573
01:14
its impact will be so enormous
18
62017
2162
01:16
that it will make the first
software revolution pale in comparison.
software revolution pale in comparison.
19
64203
3877
01:20
And that's because living software
would transform the entirety of medicine,
would transform the entirety of medicine,
20
68104
4234
01:24
agriculture and energy,
21
72362
1559
01:25
and these are sectors that dwarf
those dominated by IT.
those dominated by IT.
22
73945
3828
01:30
Imagine programmable plants
that fix nitrogen more effectively
that fix nitrogen more effectively
23
78812
4174
01:35
or resist emerging fungal pathogens,
24
83010
2905
01:37
or even programming crops
to be perennial rather than annual
to be perennial rather than annual
25
85939
3537
01:41
so you could double
your crop yields each year.
your crop yields each year.
26
89500
2268
01:43
That would transform agriculture
27
91792
2098
01:45
and how we'll keep our growing
and global population fed.
and global population fed.
28
93914
4104
01:50
Or imagine programmable immunity,
29
98794
2262
01:53
designing and harnessing molecular devices
that guide your immune system
that guide your immune system
30
101080
4238
01:57
to detect, eradicate
or even prevent disease.
or even prevent disease.
31
105342
3830
02:01
This would transform medicine
32
109196
1571
02:02
and how we'll keep our growing
and aging population healthy.
and aging population healthy.
33
110791
3489
02:07
We already have many of the tools
that will make living software a reality.
that will make living software a reality.
34
115501
4203
02:11
We can precisely edit genes with CRISPR.
35
119728
2347
02:14
We can rewrite the genetic code
one base at a time.
one base at a time.
36
122099
3083
02:17
We can even build functioning
synthetic circuits out of DNA.
synthetic circuits out of DNA.
37
125206
4436
02:22
But figuring out how and when
to wield these tools
to wield these tools
38
130428
2469
02:24
is still a process of trial and error.
39
132921
2422
02:27
It needs deep expertise,
years of specialization.
years of specialization.
40
135367
3660
02:31
And experimental protocols
are difficult to discover
are difficult to discover
41
139051
3037
02:34
and all too often, difficult to reproduce.
42
142112
2582
02:37
And, you know, we have a tendency
in biology to focus a lot on the parts,
in biology to focus a lot on the parts,
43
145256
4473
02:41
but we all know that something like flying
wouldn't be understood
wouldn't be understood
44
149753
3133
02:44
by only studying feathers.
45
152910
1339
02:46
So programming biology is not yet
as simple as programming your computer.
as simple as programming your computer.
46
154846
4521
02:51
And then to make matters worse,
47
159391
1678
02:53
living systems largely bear no resemblance
to the engineered systems
to the engineered systems
48
161093
4010
02:57
that you and I program every day.
49
165127
2096
02:59
In contrast to engineered systems,
living systems self-generate,
living systems self-generate,
50
167691
4111
03:03
they self-organize,
51
171826
1471
03:05
they operate at molecular scales.
52
173321
1687
03:07
And these molecular-level interactions
53
175032
2136
03:09
lead generally to robust
macro-scale output.
macro-scale output.
54
177192
3018
03:12
They can even self-repair.
55
180234
2720
03:16
Consider, for example,
the humble household plant,
the humble household plant,
56
184256
2994
03:19
like that one sat
on your mantelpiece at home
on your mantelpiece at home
57
187274
2187
03:21
that you keep forgetting to water.
58
189485
1787
03:23
Every day, despite your neglect,
that plant has to wake up
that plant has to wake up
59
191749
3615
03:27
and figure out how
to allocate its resources.
to allocate its resources.
60
195388
2747
03:30
Will it grow, photosynthesize,
produce seeds, or flower?
produce seeds, or flower?
61
198159
3571
03:33
And that's a decision that has to be made
at the level of the whole organism.
at the level of the whole organism.
62
201754
3939
03:37
But a plant doesn't have a brain
to figure all of that out.
to figure all of that out.
63
205717
3481
03:41
It has to make do
with the cells on its leaves.
with the cells on its leaves.
64
209222
2717
03:43
They have to respond to the environment
65
211963
1903
03:45
and make the decisions
that affect the whole plant.
that affect the whole plant.
66
213890
2649
03:48
So somehow there must be a program
running inside these cells,
running inside these cells,
67
216563
3988
03:52
a program that responds
to input signals and cues
to input signals and cues
68
220575
2727
03:55
and shapes what that cell will do.
69
223326
1940
03:57
And then those programs must operate
in a distributed way
in a distributed way
70
225679
3247
04:00
across individual cells,
71
228950
1337
04:02
so that they can coordinate
and that plant can grow and flourish.
and that plant can grow and flourish.
72
230311
4123
04:07
If we could understand
these biological programs,
these biological programs,
73
235675
3316
04:11
if we could understand
biological computation,
biological computation,
74
239015
3122
04:14
it would transform our ability
to understand how and why
to understand how and why
75
242161
3937
04:18
cells do what they do.
76
246122
1546
04:20
Because, if we understood these programs,
77
248152
1987
04:22
we could debug them when things go wrong.
78
250163
2133
04:24
Or we could learn from them how to design
the kind of synthetic circuits
the kind of synthetic circuits
79
252320
4193
04:28
that truly exploit
the computational power of biochemistry.
the computational power of biochemistry.
80
256537
4474
04:34
My passion about this idea
led me to a career in research
led me to a career in research
81
262407
3018
04:37
at the interface of maths,
computer science and biology.
computer science and biology.
82
265449
3631
04:41
And in my work, I focus on the concept
of biology as computation.
of biology as computation.
83
269104
4726
04:46
And that means asking
what do cells compute,
what do cells compute,
84
274334
3142
04:49
and how can we uncover
these biological programs?
these biological programs?
85
277500
3517
04:53
And I started to ask these questions
together with some brilliant collaborators
together with some brilliant collaborators
86
281760
3757
04:57
at Microsoft Research
and the University of Cambridge,
and the University of Cambridge,
87
285541
2571
05:00
where together we wanted to understand
88
288136
2283
05:02
the biological program
running inside a unique type of cell:
running inside a unique type of cell:
89
290443
4177
05:06
an embryonic stem cell.
90
294644
1894
05:09
These cells are unique
because they're totally naïve.
because they're totally naïve.
91
297136
3160
05:12
They can become anything they want:
92
300320
2168
05:14
a brain cell, a heart cell,
a bone cell, a lung cell,
a bone cell, a lung cell,
93
302512
2565
05:17
any adult cell type.
94
305101
1897
05:19
This naïvety, it sets them apart,
95
307022
1677
05:20
but it also ignited the imagination
of the scientific community,
of the scientific community,
96
308723
3001
05:23
who realized, if we could
tap into that potential,
tap into that potential,
97
311748
3263
05:27
we would have a powerful
tool for medicine.
tool for medicine.
98
315035
2351
05:29
If we could figure out
how these cells make the decision
how these cells make the decision
99
317917
2621
05:32
to become one cell type or another,
100
320562
2131
05:34
we might be able to harness them
101
322717
1690
05:36
to generate cells that we need
to repair diseased or damaged tissue.
to repair diseased or damaged tissue.
102
324431
4553
05:41
But realizing that vision
is not without its challenges,
is not without its challenges,
103
329794
2930
05:44
not least because these particular cells,
104
332748
2764
05:47
they emerge just six days
after conception.
after conception.
105
335536
2829
05:50
And then within a day or so, they're gone.
106
338826
2055
05:52
They have set off down the different paths
107
340905
2057
05:54
that form all the structures
and organs of your adult body.
and organs of your adult body.
108
342986
3050
05:59
But it turns out that cell fates
are a lot more plastic
are a lot more plastic
109
347770
3079
06:02
than we might have imagined.
110
350873
1413
06:04
About 13 years ago, some scientists
showed something truly revolutionary.
showed something truly revolutionary.
111
352310
4321
06:09
By inserting just a handful of genes
into an adult cell,
into an adult cell,
112
357393
4346
06:13
like one of your skin cells,
113
361763
1764
06:15
you can transform that cell
back to the naïve state.
back to the naïve state.
114
363551
3959
06:19
And it's a process that's actually
known as "reprogramming,"
known as "reprogramming,"
115
367534
3175
06:22
and it allows us to imagine
a kind of stem cell utopia,
a kind of stem cell utopia,
116
370733
3359
06:26
the ability to take a sample
of a patient's own cells,
of a patient's own cells,
117
374116
3641
06:29
transform them back to the naïve state
118
377781
2360
06:32
and use those cells to make
whatever that patient might need,
whatever that patient might need,
119
380165
3130
06:35
whether it's brain cells or heart cells.
120
383319
2075
06:38
But over the last decade or so,
121
386541
1765
06:40
figuring out how to change cell fate,
122
388330
3044
06:43
it's still a process of trial and error.
123
391398
2152
06:45
Even in cases where we've uncovered
successful experimental protocols,
successful experimental protocols,
124
393911
4508
06:50
they're still inefficient,
125
398443
1467
06:51
and we lack a fundamental understanding
of how and why they work.
of how and why they work.
126
399934
4238
06:56
If you figured out how to change
a stem cell into a heart cell,
a stem cell into a heart cell,
127
404650
3005
06:59
that hasn't got any way of telling you
how to change a stem cell
how to change a stem cell
128
407679
3089
07:02
into a brain cell.
129
410792
1201
07:04
So we wanted to understand
the biological program
the biological program
130
412633
2931
07:07
running inside an embryonic stem cell,
131
415588
2447
07:10
and understanding the computation
performed by a living system
performed by a living system
132
418059
3506
07:13
starts with asking
a devastatingly simple question:
a devastatingly simple question:
133
421589
4253
07:17
What is it that system actually has to do?
134
425866
3356
07:21
Now, computer science actually
has a set of strategies
has a set of strategies
135
429838
2850
07:24
for dealing with what it is the software
and hardware are meant to do.
and hardware are meant to do.
136
432712
3827
07:28
When you write a program,
you code a piece of software,
you code a piece of software,
137
436563
2660
07:31
you want that software to run correctly.
138
439247
2000
07:33
You want performance, functionality.
139
441271
1790
07:35
You want to prevent bugs.
140
443085
1217
07:36
They can cost you a lot.
141
444326
1308
07:38
So when a developer writes a program,
142
446168
1842
07:40
they could write down
a set of specifications.
a set of specifications.
143
448034
2270
07:42
These are what your program should do.
144
450328
1871
07:44
Maybe it should compare
the size of two numbers
the size of two numbers
145
452223
2268
07:46
or order numbers by increasing size.
146
454515
1792
07:49
Technology exists that allows us
automatically to check
automatically to check
147
457037
4695
07:53
whether our specifications are satisfied,
148
461756
2378
07:56
whether that program
does what it should do.
does what it should do.
149
464158
2633
07:59
And so our idea was that in the same way,
150
467266
2856
08:02
experimental observations,
things we measure in the lab,
things we measure in the lab,
151
470146
3068
08:05
they correspond to specifications
of what the biological program should do.
of what the biological program should do.
152
473238
5033
08:10
So we just needed to figure out a way
153
478769
1876
08:12
to encode this new type of specification.
154
480669
3183
08:16
So let's say you've been busy in the lab
and you've been measuring your genes
and you've been measuring your genes
155
484594
3654
08:20
and you've found that if Gene A is active,
156
488272
2436
08:22
then Gene B or Gene C seems to be active.
157
490732
3388
08:26
We can write that observation down
as a mathematical expression
as a mathematical expression
158
494678
3582
08:30
if we can use the language of logic:
159
498284
2373
08:33
If A, then B or C.
160
501125
2328
08:36
Now, this is a very simple example, OK.
161
504242
2454
08:38
It's just to illustrate the point.
162
506720
1743
08:40
We can encode truly rich expressions
163
508487
2924
08:43
that actually capture the behavior
of multiple genes or proteins over time
of multiple genes or proteins over time
164
511435
4153
08:47
across multiple different experiments.
165
515612
2536
08:50
And so by translating our observations
166
518521
2626
08:53
into mathematical expression in this way,
167
521171
1993
08:55
it becomes possible to test whether
or not those observations can emerge
or not those observations can emerge
168
523188
5098
09:00
from a program of genetic interactions.
169
528310
3054
09:04
And we developed a tool to do just this.
170
532063
2556
09:06
We were able to use this tool
to encode observations
to encode observations
171
534643
2882
09:09
as mathematical expressions,
172
537549
1407
09:10
and then that tool would allow us
to uncover the genetic program
to uncover the genetic program
173
538980
3610
09:14
that could explain them all.
174
542614
1538
09:17
And we then apply this approach
175
545481
2280
09:19
to uncover the genetic program
running inside embryonic stem cells
running inside embryonic stem cells
176
547785
4083
09:23
to see if we could understand
how to induce that naïve state.
how to induce that naïve state.
177
551892
4189
09:28
And this tool was actually built
178
556105
1952
09:30
on a solver that's deployed
routinely around the world
routinely around the world
179
558081
2652
09:32
for conventional software verification.
180
560757
2269
09:35
So we started with a set
of nearly 50 different specifications
of nearly 50 different specifications
181
563630
3691
09:39
that we generated from experimental
observations of embryonic stem cells.
observations of embryonic stem cells.
182
567345
4506
09:43
And by encoding these
observations in this tool,
observations in this tool,
183
571875
2636
09:46
we were able to uncover
the first molecular program
the first molecular program
184
574535
3185
09:49
that could explain all of them.
185
577744
1961
09:52
Now, that's kind of a feat
in and of itself, right?
in and of itself, right?
186
580309
2513
09:54
Being able to reconcile
all of these different observations
all of these different observations
187
582846
2902
09:57
is not the kind of thing
you can do on the back of an envelope,
you can do on the back of an envelope,
188
585772
3067
10:00
even if you have a really big envelope.
189
588863
2648
10:04
Because we've got
this kind of understanding,
this kind of understanding,
190
592190
2158
10:06
we could go one step further.
191
594372
1462
10:07
We could use this program to predict
what this cell might do
what this cell might do
192
595858
3371
10:11
in conditions we hadn't yet tested.
193
599253
2176
10:13
We could probe the program in silico.
194
601453
2401
10:16
And so we did just that:
195
604735
1247
10:18
we generated predictions
that we tested in the lab,
that we tested in the lab,
196
606006
3180
10:21
and we found that this program
was highly predictive.
was highly predictive.
197
609210
3032
10:24
It told us how we could
accelerate progress
accelerate progress
198
612266
2625
10:26
back to the naïve state
quickly and efficiently.
quickly and efficiently.
199
614915
3060
10:29
It told us which genes
to target to do that,
to target to do that,
200
617999
2570
10:32
which genes might even
hinder that process.
hinder that process.
201
620593
2624
10:35
We even found the program predicted
the order in which genes would switch on.
the order in which genes would switch on.
202
623241
4990
10:40
So this approach really allowed us
to uncover the dynamics
to uncover the dynamics
203
628980
3140
10:44
of what the cells are doing.
204
632144
2402
10:47
What we've developed, it's not a method
that's specific to stem cell biology.
that's specific to stem cell biology.
205
635728
3642
10:51
Rather, it allows us to make sense
of the computation
of the computation
206
639394
2684
10:54
being carried out by the cell
207
642102
1685
10:55
in the context of genetic interactions.
208
643811
2831
10:58
So really, it's just one building block.
209
646666
2288
11:00
The field urgently needs
to develop new approaches
to develop new approaches
210
648978
2685
11:03
to understand biological
computation more broadly
computation more broadly
211
651687
2695
11:06
and at different levels,
212
654406
1367
11:07
from DNA right through
to the flow of information between cells.
to the flow of information between cells.
213
655797
4129
11:11
Only this kind of
transformative understanding
transformative understanding
214
659950
2797
11:14
will enable us to harness biology
in ways that are predictable and reliable.
in ways that are predictable and reliable.
215
662771
4986
11:21
But to program biology,
we will also need to develop
we will also need to develop
216
669029
3042
11:24
the kinds of tools and languages
217
672095
1995
11:26
that allow both experimentalists
and computational scientists
and computational scientists
218
674114
3408
11:29
to design biological function
219
677546
2497
11:32
and have those designs compile down
to the machine code of the cell,
to the machine code of the cell,
220
680067
3505
11:35
its biochemistry,
221
683596
1181
11:36
so that we could then
build those structures.
build those structures.
222
684801
2484
11:39
Now, that's something akin
to a living software compiler,
to a living software compiler,
223
687309
3673
11:43
and I'm proud to be
part of a team at Microsoft
part of a team at Microsoft
224
691006
2216
11:45
that's working to develop one.
225
693246
1652
11:47
Though to say it's a grand challenge
is kind of an understatement,
is kind of an understatement,
226
695366
3226
11:50
but if it's realized,
227
698616
1173
11:51
it would be the final bridge
between software and wetware.
between software and wetware.
228
699813
3709
11:57
More broadly, though, programming biology
is only going to be possible
is only going to be possible
229
705006
3415
12:00
if we can transform the field
into being truly interdisciplinary.
into being truly interdisciplinary.
230
708445
4279
12:04
It needs us to bridge
the physical and the life sciences,
the physical and the life sciences,
231
712748
2952
12:07
and scientists from
each of these disciplines
each of these disciplines
232
715724
2267
12:10
need to be able to work together
with common languages
with common languages
233
718015
2731
12:12
and to have shared scientific questions.
234
720770
2719
12:16
In the long term, it's worth remembering
that many of the giant software companies
that many of the giant software companies
235
724757
3993
12:20
and the technology
that you and I work with every day
that you and I work with every day
236
728774
2492
12:23
could hardly have been imagined
237
731290
1503
12:24
at the time we first started
programming on silicon microchips.
programming on silicon microchips.
238
732817
3605
12:28
And if we start now to think about
the potential for technology
the potential for technology
239
736446
3031
12:31
enabled by computational biology,
240
739501
2426
12:33
we'll see some of the steps
that we need to take along the way
that we need to take along the way
241
741951
2935
12:36
to make that a reality.
242
744910
1433
12:39
Now, there is the sobering thought
that this kind of technology
that this kind of technology
243
747231
3082
12:42
could be open to misuse.
244
750337
1777
12:44
If we're willing to talk
about the potential
about the potential
245
752138
2163
12:46
for programming immune cells,
246
754325
1436
12:47
we should also be thinking
about the potential of bacteria
about the potential of bacteria
247
755785
3188
12:50
engineered to evade them.
248
758997
1661
12:52
There might be people willing to do that.
249
760682
2087
12:55
Now, one reassuring thought in this
250
763506
1722
12:57
is that -- well, less so
for the scientists --
for the scientists --
251
765252
2289
12:59
is that biology is
a fragile thing to work with.
a fragile thing to work with.
252
767565
3269
13:02
So programming biology
is not going to be something
is not going to be something
253
770858
2412
13:05
you'll be doing in your garden shed.
254
773294
1848
13:07
But because we're at the outset of this,
255
775642
2080
13:09
we can move forward
with our eyes wide open.
with our eyes wide open.
256
777746
2583
13:12
We can ask the difficult
questions up front,
questions up front,
257
780353
2324
13:14
we can put in place
the necessary safeguards
the necessary safeguards
258
782701
3040
13:17
and, as part of that,
we'll have to think about our ethics.
we'll have to think about our ethics.
259
785765
2797
13:20
We'll have to think about putting bounds
on the implementation
on the implementation
260
788586
3172
13:23
of biological function.
261
791782
1498
13:25
So as part of this, research in bioethics
will have to be a priority.
will have to be a priority.
262
793604
3715
13:29
It can't be relegated to second place
263
797343
2407
13:31
in the excitement
of scientific innovation.
of scientific innovation.
264
799774
2514
13:35
But the ultimate prize,
the ultimate destination on this journey,
the ultimate destination on this journey,
265
803154
3474
13:38
would be breakthrough applications
and breakthrough industries
and breakthrough industries
266
806652
3444
13:42
in areas from agriculture and medicine
to energy and materials
to energy and materials
267
810120
3444
13:45
and even computing itself.
268
813588
2261
13:48
Imagine, one day we could be powering
the planet sustainably
the planet sustainably
269
816490
3148
13:51
on the ultimate green energy
270
819662
1859
13:53
if we could mimic something
that plants figured out millennia ago:
that plants figured out millennia ago:
271
821545
3943
13:57
how to harness the sun's energy
with an efficiency that is unparalleled
with an efficiency that is unparalleled
272
825512
3771
14:01
by our current solar cells.
273
829307
1856
14:03
If we understood that program
of quantum interactions
of quantum interactions
274
831695
2601
14:06
that allow plants to absorb
sunlight so efficiently,
sunlight so efficiently,
275
834320
3264
14:09
we might be able to translate that
into building synthetic DNA circuits
into building synthetic DNA circuits
276
837608
3944
14:13
that offer the material
for better solar cells.
for better solar cells.
277
841576
2913
14:17
There are teams and scientists working
on the fundamentals of this right now,
on the fundamentals of this right now,
278
845349
3693
14:21
so perhaps if it got the right attention
and the right investment,
and the right investment,
279
849066
3243
14:24
it could be realized in 10 or 15 years.
280
852333
2280
14:27
So we are at the beginning
of a technological revolution.
of a technological revolution.
281
855457
3197
14:31
Understanding this ancient type
of biological computation
of biological computation
282
859067
3221
14:34
is the critical first step.
283
862312
2132
14:36
And if we can realize this,
284
864468
1317
14:37
we would enter in the era
of an operating system
of an operating system
285
865809
2842
14:40
that runs living software.
286
868675
1905
14:42
Thank you very much.
287
870604
1166
14:43
(Applause)
288
871794
2690
ABOUT THE SPEAKER
Sara-Jane Dunn - Computational biologistSara-Jane Dunn is a scientist working at the interface between biology and computation, using mathematics and computational analysis to make sense of how living systems process information.
Why you should listen
Sara-Jane Dunn started her career as a mathematician, finding a niche in "mathematical biology" -- a field where mathematical models and simulations provide insights into the behavior and development of living systems. For Dunn, this research opened up new ways to explore old problems, from understanding how diseases spread to how cancer evolves. To this end, she created computational models of the gut that could be used to test hypotheses about the earliest stages of colorectal cancer.
At Microsoft Research, Dunn collaborates with experimenters in embryonic stem cell biology and seeks to uncover the biological program that governs how these unique cells are able to generate all of the different cell types of your adult body. This work could, in time, "unlock" biology and make cells programmable, which could fundamentally transform medicine, agriculture and even how we power the planet.
Sara-Jane Dunn | Speaker | TED.com