M. J. Dunham, S. D. Emr, F. Hughson, J. Hurley, K. Murakami,
N. Nakamura, E. Nogales, R. Schekman, S. Shan, S. Showman,
K. Sugasawa, and S. Suzuki for their correspondence and biological
expertise. We thank S. Burley, B. Vallat, and J. Westbrook at
the RCSB Protein Data Bank and T. Schwede, G. Tauriello,
A. Waterhouse, and S. Bienert at SWISS-MODEL for hosting our
model structures at the ModelArchive.Funding:This work was
supported by Microsoft (M.B., D.B., and Azure compute
time and expertise), Amgen (D.B. and I.R.H.), Southwestern
Medical Foundation (J.P. and Q.C.), the Washington Research
Foundation (M.B. and Q.C.), the Howard Hughes Medical Institute
(D.B., S.B., S.K., and generous compute time on Janelia), National
Science Foundation (NSF) Cyberinfrastructure for Biological
Research (CIBR, Award DBI 1937533 to D.B. and I.A.), CPRIT
training grant (RP210041 to J.Z.), UK Medical Research Council
(MRC_UP_1201/10 to E.A.M.), HHMI Gilliam Fellowship (D.J.B.),
the Deutsche Forschungsgemeinschaft (KI-562/11-1 and KI-562/7-1 to
C.K.), NIH/NIGMS (R21AI156595 to S.O., R35GM136258 to J.C.F.,
R35NS097333 to J.R., R35GM118026 and R01CA221858 to E.C.G.),
HHMI fellowship of the Damon Runyon Cancer Research
Foundation (DRG2273-16 to S.B. and DRG2389-20 to K.L.),
AIRC investigator and the European Research Council Consolidator
(IG23710 and 682190 to D.Br.), and the Defense Threat Reduction
Agency (HDTRA1-21-1-0007 to D.B.). We also thank The National
Energy Research Scientific Computing Center (NERSC) for
providing computing time (project m3962 at NERSC).Author
contributions:Q.C. and D.B. conceived the research; J.P. and
Q.C. prepared the sequence alignments used in the screen; M.B.
implemented the RoseTTAFold pipeline; M.B. and S.O. repurposed
AlphaFold for complex modeling; J.P., J.Z., and Q.C. designed
the PPI screening procedure; I.R.H., M.B., I.A., and Q.C. carried out
the screen; I.R.H., A.K., and Q.C. analyzed and presented the
results; I.R.H., A.K., Q.C., and D.B. coordinated the collaborative
efforts; T.J.N., S.B., S.R.B., V.G.S., X.H.L., K.L., Z.Z., D.J.B., U.R.,
J.K., I.S.F., B.S., D.B., J.R., C.K., E.C.G., S.B., S.K., E.A.M., J.C.F., and
T.L.H. provided biological insights on specific examples; Q.C.
and D.B. drafted the manuscript and all other authors contributed
to the description of specific examples; all authors discussed
the results and commented on the manuscript.Competing
interests:The authors declare that they have no competing
interests.Data and materials availability:Structures of highly
confident pairs with accompanying pMSAs and metadata are
available at the ModelArchive:https://modelarchive.org/doi/
10.5452/ma-bak-cepc. RoseTTAFold two-track version isavailable athttps://github.com/RosettaCommons/RoseTTAFold
or Zenodo ( 101 ). AlphaFold was fetched fromhttps://github.com/
deepmind/alphafoldon 16 July 2021 (v2.0.0). Code for a GPU
implementation of DCA and the modifications to the AlphaFold
predictions script are provided in supplementary materials
and methods.SUPPLEMENTARY MATERIALS
science.org/doi/10.1126/science.amb4805
Materials and Methods
Supplementary Text
Figs. S1 to S27
Tables S1 to S5
References ( 102 Ð 116 )
Descriptions of all predicted protein-protein interactions20 September 2021; accepted 2 November 2021
Published online 11 November 2021
10.1126/science.abm4805Humphreyset al.,Science 374 , eabm4805 (2021) 10 December 2021 12 of 12
RESEARCH | RESEARCH ARTICLE