The DOI matches its Crossref record; 1 of the 1 references it cites were found in a registry or an index.
Learning the local landscape of protein structures with convolutional neural networks
Open accessPreprint — published as 10.1007/s10867-021-09593-6
Abstract
The fundamental problem of protein biochemistry is to predict protein structure from amino acid sequence. The inverse problem, predicting either entire sequences or individual mutations that are consistent with a given protein structure, has received much less attention even though it has important applications in both protein engineering and evolutionary biology. Here, we ask whether 3D convolutional neural networks (3D CNNs) can learn the local fitness landscape of protein structure to reliably predict either the wild-type amino acid or the consensus in a multiple sequence alignment from the local structural context surrounding a site of interest. We find that the network can predict wild type with good accuracy, and that network confidence is a reliable measure of whether a given prediction is likely going to be correct or not. Predictions of consensus are less accurate, and are primarily driven by whether or not the consensus matches the wild type. Our work suggests that high-confidence mis-predictions of the wild type may identify sites that are primed for mutation and likely targets for protein engineering.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Highly accurate protein structure prediction with AlphaFold
Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., et al. (2021) · Nature · doi:10.1038/s41586-021-03819-2 · Cited by 43,764
Flagged first, then by citations
Research works printed but not linked: 36
- 2016, Detection and sequence/structure mapping of biophysical constraints to protein variation in saturated mutational libraries and protein sequence alignments with a dedicated server, BMC Bioinf, 17, 242, 10.1186/s12859-016-1124-4 · DOI
- 1995, Controlling the false discovery rate: A practical and powerful approach to multiple testing, J R Stat Soc Series B Stat Methodol J R STAT SOC B, 57, 289, 10.1111/j.2517-6161.1995.tb02031.x · DOI · Paper page
- Bisardi, M. , Rodriguez-Rivas, J. , Zamponi, F. , Weigt, M. : Modeling sequence-space exploration and emergence of epistatic signals in protein evolution (2021) · DOI · Paper page
- 10.1093/nar/gkm276 · DOI
- 10.1073/pnas.0605504103 · DOI
- 10.1038/nrg.2015.18 · DOI
- 10.1146/annurev-biophys-070816-033819 · DOI
- 10.1126/science.1749933 · DOI · Paper page
- 2016, The tangled bank of amino acids, Protein Sci, 25, 1354, 10.1002/pro.2930 · DOI
- 2017, Sequence entropy of folding and the absolute rate of amino acid substitutions, Nature Ecol. Evol, 1, 1923, 10.1038/s41559-017-0338-9 · DOI
- 2019, Learning from protein fitness landscapes: a review of mutability, epistasis, and evolution, Curr. Opin. Syst. Biol, 14, 25, 10.1016/j.coisb.2019.02.006 · DOI
- 10.1073/pnas.89.22.10915 · DOI · Paper page
- Huang, T.T. , d. V. Marcos, M.L. , Hwang, J.K. , Echave, J. : A mechanistic stress model of protein evolution accounts for site-specific evolutionary rates and their relationship with packing density and flexibility. BMC Evol. Biol. 14(2014) · DOI
- 10.1371/journal.pbio.1002469 · DOI · Paper page
- 10.1534/genetics.118.300699 · DOI
- 2020, Site-specific amino acid distributions follow a universal shape, J. Mol. Evol, 88, 731, 10.1007/s00239-020-09976-8 · DOI
- 2011, PSICOV: precise structural contact prediction using sparse inverse covariance estimation on large multiple sequence alignments, Bioinformatics, 28, 184
- 10.1126/science.aaa0769 · DOI
- 10.1093/molbev/msn067 · DOI · Paper page
- 2015, Too packed to change: side-chain packing and site-specific substitution rates in protein evolution, PeerJ, 3, e911, 10.7717/peerj.911 · DOI
- 10.1006/jmbi.1999.2911 · DOI
- 2016, FreeSASA: an open source C library for solvent accessible surface area calculations [version 1; peer review: 2 approved], F1000Research, 5, 189, 10.12688/f1000research.7931.1 · DOI
- 2016, Long-range epistasis mediated by structural change in a model of ligand binding proteins, PLoS ONE, 11, e0166739, 10.1371/journal.pone.0166739 · DOI
- 10.1371/journal.pgen.1008465 · DOI
- 10.1073/pnas.1120084109 · DOI
- 2021, Expanding functional protein sequence spaces using generative adversarial networks, Nat. Mach. Intell, 3, 324, 10.1038/s42256-021-00310-5 · DOI · Paper page
- Rives, A. , Meier, J. , Sercu, T. , Goyal, S. , Lin, Z. , Liu, J. , Guo, D. , Ott, M. , Zitnick, C.L. , Ma, J. , Fergus, R. : Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences. Proc. Natl. Acad. Sci. U.S.A. 118(15) (2021) · DOI · Paper page
- 10.1073/pnas.1412933112 · DOI
- 2018, Nature of long-range evolutionary constraint in enzymes: insights from comparison to pseudoenzymes with similar structures, Mol. Biol. Evol, 35, 2597, 10.1093/molbev/msy177 · DOI
- 2020, Discovery of novel gain-of-function mutations guided by structure-based deep learning, ACS Synth. Biol, 9, 2927, 10.1021/acssynbio.0c00345 · DOI
- 10.1021/j100058a043 · DOI
- 10.1093/molbev/msy210 · DOI
- 2017, 3D deep convolutional neural networks for amino acid environment similarity analysis, BMC Bioinf, 18, 302, 10.1186/s12859-017-1702-0 · DOI · Paper page
- 10.1093/oxfordjournals.molbev.a003851 · DOI · Paper page
- 2019, Welcome to the tidyverse, J. Open Source Softw, 4, 1686, 10.21105/joss.01686 · DOI · Paper page
- 2020, Deep dive into machine learning models for protein engineering, J. Chem. Inf. Model, 60, 2773, 10.1021/acs.jcim.0c00073 · DOI
Metadata
- Authors
- Kulikova, A. V., Diaz, D. J., Loy, J. M., Ellington, A. D., & Wilke, C. O.
- Affiliations
- The University of Texas at Austin (US)
- Corresponding author
- Wilke, Claus O.
- Venue
- Journal of Biological Physics
- Publisher
- openRxiv
- Published
- 2021-08-19
- Volume
- 47
- Issue
- 4
- Pages
- 435-454
- ISSN
- 0092-0606 (ISSN-L)
- PMC
- PMC8603988
- OpenAlex
- W3214430077
- Cited by
- 23
- References
- 1
- Citations by year
- 2026: 1 · 2025: 4 · 2024: 12 · 2023: 4 · 2022: 2
- FWCI
- 1.41
- Type
- journal article
- Language
- en
- Open access
- green
- doi.org
- Licence
- cc-by
- Locations
- Journal of Biological Physics (publishedVersion)
- bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by)
- PubMed Central (publishedVersion)
- PubMed (publishedVersion)
- Funders
- U.S. Department of Defense (HDTRA12010011)
- Welch Foundation (F-1654)
- Advanced Micro Devices
- National Institutes of Health
- Defense Threat Reduction Agency (HDTRA12010011)
- National Institute of Allergy and Infectious Diseases (R01AI148419)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Machine Learning in Bioinformatics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Structural Biology › Enzyme Structure and Function
- Keywords
- protein structure prediction, inverse problems, 3D convolutional neural networks, prediction confidence, protein engineering
- MeSH terms
- Amino Acid Sequence, Amino Acids, Proteins, Neural Networks, Computer
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
