262 16 JULY 2021 • VOL 373 ISSUE 6552 sciencemag.org SCIENCEIMAGE: IAN HAYDON/INSTITUTE FOR PROTEIN DESIGNP
roteins are the minions of life, work-
ing alone or together to build, manage,
fuel, protect, and eventually destroy
cells. To function, these long chains of
amino acids twist and fold and inter-
twine into complex shapes that can be
slow, even impossible, to decipher. Scientists
have dreamed of simply predicting a pro-
tein’s shape from its amino acid sequence—
an ability that would open a world of insights
into the workings of life. “This problem has
been around for 50 years; lots of people have
broken their head on it,” says John Moult, a
structural biologist at the University of Mary-
land, Shady Grove. But a practical solution is
in their grasp.
Several months ago, in a result hailed as
a turning point, computational biologists
showed that artificial intelligence (AI) could
accurately predict protein shapes. Now, David
Baker and Minkyung Baek at the University
of Washington, Seattle, and their colleagues
have made AI-based structure prediction
more powerful and accessible. Their method,
described online in Science this week, works
on not just simple proteins, but also com-plexes of proteins, and its creators have made
their computer code freely available.
Since the method was posted online last
month, the team has used it to model more
than 4500 protein sequences submitted by
other researchers. Savvas Savvides, a struc-
tural biologist at Ghent University, had
tried six times to model a problematic pro-
tein. He says Baker’s and Baek’s program,
called RoseTTAFold, “paved the way to a
structure solution.”
In fall of 2020, DeepMind, a U.K.-based
AI company owned by Google, wowed the
field with its structure predictions in a bien-
nial competition (Science, 4 December 2020,
p. 1144). Called Critical Assessment of Protein
Structure Prediction (CASP), the competition
uses structures newly determined using labo-
rious lab techniques such as x-ray crystallo-
graphy as benchmarks. DeepMind’s program,
AlphaFold2, did “really extraordinary things
[predicting] protein structures with atomic
accuracy,” says Moult, who organizes CASP.
But for many structural biologists,
AlphaFold2 was a tease: “Incredibly exciting
but also very frustrating,” says David Agard,
a structural biophysicist at the University of
California, San Francisco. DeepMind has yetto publish its method and computer code
for others to take advantage of. In mid-June,
3 days after the Baker lab posted its RoseTTA-
Fold preprint, Demis Hassabis, DeepMind’s
CEO, tweeted that AlphaFold2’s details were
under review at a publication and the com-
pany would provide “broad free access to
AlphaFold for the scientific community.”
DeepMind’s 30-minute presentation at
CASP was enough to inspire Baek to develop
her own approach. Like AlphaFold2, it uses
AI’s ability to discern patterns in vast data-
bases of examples, generating ever more in-
formed and accurate iterations as it learns.
When given a new protein to model, Rose-
TTAFold proceeds along multiple “tracks.”
One compares the protein’s amino acid
sequence with all similar sequences in pro-
tein databases. Another predicts pairwise
interactions between amino acids within the
protein, and a third compiles the putative 3D
structure. The program bounces among the
tracks to refine the model, using the output
of each one to update the others.
DeepMind’s approach, although still un-
der wraps, involves just two tracks, Baek and
others believe. Gira Bhabha, a cell and struc-
tural biologist at New York University SchoolIN DEPTH
A new artificial intelligence program readily predicts the structure of protein complexes, such as the immune signal interleukin-12 (blue) bound to its receptor.By Elizabeth PennisiSTRUCTURAL BIOLOGYProtein structure prediction now easier, faster
AI approach is accessible to all structural biology, drug discovery researchers
0716NewsInDepth.indd 262 7/13/21 5:52 PM