0:00

Today in Science from Wired This

0:06

episode is brought to you by ShipStation. You

0:09

know folks, I'm all about working smarter, not

0:11

harder. Running a successful e-commerce

0:13

business can be really convoluted and honestly,

0:15

you have enough going on, so I

0:17

want to introduce you guys to ShipStation.

0:20

I love using ShipStation because it makes

0:22

it so much easier to automate all

0:25

my shipping tasks and I can manage

0:27

all my orders in one simple dashboard.

0:29

Work less and ship more with

0:32

ShipStation, the innovative tool that helps

0:34

turn your shipping challenges into opportunities

0:36

for growth. Go to shipstation.com

0:38

and use code TECHNUCE to

0:41

sign up for your free

0:43

60-day trial. That's shipstation.com code

0:45

TECHNUCE. And now a

0:47

next-level moment brought to you by AT&T Business.

0:49

Imagine your business is running like clockwork and

0:51

everything is going great. Wow, everything

0:53

is going great. Then suddenly there's

0:55

a cyber threat. But with real-time threat detection

0:57

and response built in, your network stops the

1:00

threat before you even know it's there. Wow,

1:02

everything is still going great. That's

1:04

the confidence you have with the only

1:06

network with comprehensive built-in security controls. Get

1:08

AT&T Dynamic Defense, exclusively available in select

1:10

areas with AT&T-dedicated internet and help stop

1:13

cyber threats in their tracks. With

1:17

AI tools, scientists can crack the code of

1:19

life. Google's AI research lab,

1:21

DeepMind, is steadily building knowledge of how

1:23

genes and their products work inside the

1:25

body. And how and why they

1:27

sometimes go wrong. Bayjuo

1:30

Muderos. In

1:33

2021, AI research lab DeepMind

1:35

announced the development of its

1:37

first digital biology neural network,

1:40

AlphaFold. The model

1:42

was capable of accurately predicting the

1:44

3D structure of proteins, which determines

1:46

the functions that these molecules play.

1:49

We're just floating bags of water moving

1:52

around, says Pushmeat Coley, VP

1:54

of research at DeepMind. What

1:56

makes us special are proteins, the building blocks

1:58

of life, how they interact with the world. with

2:00

each other is what makes the magic of life

2:02

happen. AlphaFold was considered by

2:04

the journal Science as the breakthrough of the year in

2:06

2021. In 2022, it was the most cited research paper

2:12

in AI. People have been

2:14

on protein structures for many decades and were

2:16

not able to make that much progress, Coley

2:18

says. Then came AI. DeepMind

2:21

also released the AlphaFold protein structure

2:23

database, which contained the protein structures

2:26

of almost every organism whose genome

2:28

has been sequenced, making it freely

2:30

available to scientists worldwide. More

2:33

than 1.7 million researchers in 190

2:36

countries have used it for research ranging

2:39

from the design of plastic-eating enzymes to

2:42

the development of more effective

2:44

malaria vaccines. A

2:46

quarter of the research involving

2:48

AlphaFold was dedicated to the

2:50

understanding of cancer, COVID-19, and

2:53

neurodegenerative diseases like Parkinson's and

2:55

Alzheimer's. Last year,

2:57

DeepMind released its next generation of

2:59

AlphaFold, which extended its structure prediction

3:02

algorithm to biomolecules like nucleic acids

3:04

and ligands. It

3:06

has democratized science research, Coley

3:09

says. Scientists working in a

3:11

developing country on a neglected tropical disease did

3:13

not have access to the funds to get

3:16

the structure of a protein computed. Now,

3:18

at the click of a button, they can

3:20

go to the AlphaFold database and get these

3:22

predictions for free. For

3:24

instance, one of DeepMind's early

3:26

partners, the Drugs for Neglected

3:28

Diseases Initiative, used AlphaFold to

3:30

develop medicine for diseases that

3:32

affect millions, such as sleeping

3:34

sickness, chagas disease, and leishmaniasis,

3:37

yet received comparatively little research.

3:40

DeepMind's latest breakthrough is called

3:42

AlphaMiscense. The model categorizes the

3:44

so-called Miscense mutations, genetic alterations

3:46

that can result in different

3:48

amino acids being produced at

3:50

particular positions in proteins. Such

3:53

mutations can alter the function of the

3:55

protein itself, and AlphaMiscense attributes

3:57

a likelihood score for that mutation.

4:00

being either pathogenic or benign. Understanding

4:03

and predicting those effects is crucial

4:05

for the discovery of rare genetic

4:07

diseases, Coley says. The

4:10

algorithm, which was released last year, has classified

4:12

around 89% of all possible human missense. Before,

4:16

only 0.1% of all possible

4:19

variants had been clinically classified

4:21

by researchers. This

4:23

is just the beginning, Coley says. Ultimately,

4:26

he believes AI could eventually lead

4:28

to the creation of a virtual

4:30

cell that could radically accelerate biomedical

4:33

research, enabling biology to be explored

4:35

in silico rather than in real-world

4:37

laboratories. With AI and

4:39

machine learning, we finally have the tools

4:41

to comprehend this very sophisticated system that

4:43

we call life. Thanks

4:46

for listening to WIRED. My name is

4:48

Zeke Robison, and for more stories like this one,

4:50

visit us at wired.com. Like

4:53

what you learned? Subscribe everywhere

4:55

you listen to podcasts. And

4:58

get more science news at

5:00

wired.com/science.