Protein structure prediction using multiple deep neural networks in the 13th Critical Assessment of Protein Structure Prediction (CASP13)

We describe AlphaFold, the protein structure prediction system that was entered by the group A7D in CASP13. Submissions were made by three free‐modeling (FM) methods which combine the predictions of three neural networks. All three systems were guided by predictions of distances between pairs of residues produced by a neural network. Two systems assembled fragments produced by a generative neural network, one using scores from a network trained to regress GDT_TS. The third system shows that simple gradient descent on a properly constructed potential is able to perform on par with more expensive traditional search techniques and without requiring domain segmentation. In the CASP13 FM assessors' ranking by summed z‐scores, this system scored highest with 68.3 vs 48.2 for the next closest group (an average GDT_TS of 61.4). The system produced high‐accuracy structures (with GDT_TS scores of 70 or higher) for 11 out of 43 FM domains. Despite not explicitly using template information, the results in the template category were comparable to the best performing template‐based methods.

[1]  N. Metropolis,et al.  Equation of State Calculations by Fast Computing Machines , 1953, Resonance.

[2]  W. K. Hastings,et al.  Monte Carlo Sampling Methods Using Markov Chains and Their Applications , 1970 .

[3]  C. D. Gelatt,et al.  Optimization by Simulated Annealing , 1983, Science.

[4]  Jorge Nocedal,et al.  On the limited memory BFGS method for large scale optimization , 1989, Math. Program..

[5]  W. Taylor,et al.  Estimating polypeptideα-carbon distances from multiple sequence alignments , 1995 .

[6]  Thomas L. Madden,et al.  Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. , 1997, Nucleic acids research.

[7]  C Kooperberg,et al.  Assembly of protein tertiary structures from fragments with similar local sequences using simulated annealing and Bayesian scoring functions. , 1997, Journal of molecular biology.

[8]  C Venclovas,et al.  Processing and analysis of CASP3 protein structure predictions , 1999, Proteins.

[9]  T. N. Bhat,et al.  The Protein Data Bank , 2000, Nucleic Acids Res..

[10]  Stephen H. Bryant,et al.  Domain size distributions can predict domain boundaries , 2000, Bioinform..

[11]  David C. Jones.  Predicting novel protein folds by using FRAGFOLD , 2001, Proteins.

[12]  Johannes Söding,et al.  The HHpred interactive server for protein homology detection and structure prediction , 2005, Nucleic Acids Res..

[13]  Jesper Ferkinghoff-Borg,et al.  A generative, probabilistic model of local protein structure , 2008, Proceedings of the National Academy of Sciences.

[14]  David Baker,et al.  Macromolecular modeling with rosetta. , 2008, Annual review of biochemistry.

[15]  Feng Zhao,et al.  Fragment-free approach to protein folding using conditional neural fields , 2010, Bioinform..

[16]  C. Sander,et al.  Direct-coupling analysis of residue coevolution captures native contacts across many protein families , 2011, Proceedings of the National Academy of Sciences.

[17]  A. Biegert,et al.  HHblits: lightning-fast iterative protein sequence searching by HMM-HMM alignment , 2011, Nature Methods.

[18]  Jinbo Xu,et al.  A position-specific distance-dependent statistical potential for protein structure and functional study. , 2012, Structure.

[19]  E. Aurell,et al.  Improved contact prediction in proteins: using pseudolikelihoods to infer Potts models. , 2012, Physical review. E, Statistical, nonlinear, and soft matter physics.

[20]  Kam Y. J. Zhang,et al.  Improving fragment quality for de novo structure prediction , 2014, Proteins.

[21]  Daan Wierstra,et al.  Stochastic Backpropagation and Approximate Inference in Deep Generative Models , 2014, ICML.

[22]  D. Baker,et al.  Relaxation of backbone bond geometry improves protein energy landscape modeling , 2014, Protein science : a publication of the Protein Society.

[23]  Matthew J. O’Meara,et al.  Combined covalent-electrostatic model of hydrogen bonding improves structure prediction with Rosetta. , 2015, Journal of chemical theory and computation.

[24]  David T. Jones,et al.  MetaPSICOV: combining coevolution methods for accurate prediction of contacts and long range hydrogen bonding in proteins , 2014, Bioinform..

[25]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[26]  Daan Wierstra,et al.  Towards Conceptual Compression , 2016, NIPS.

[27]  Max Jaderberg,et al.  Population Based Training of Neural Networks , 2017, ArXiv.

[28]  David A. Lee,et al.  CATH: an expanded resource to predict protein function through structure and sequence , 2016, Nucleic Acids Res..

[29]  Demis Hassabis,et al.  Mastering the game of Go without human knowledge , 2017, Nature.

[30]  Zhen Li,et al.  Accurate De Novo Prediction of Protein Contact Map by Ultra-Deep Learning Model , 2016, bioRxiv.

[31]  Yang Zhang,et al.  Template‐based and free modeling of I‐TASSER and QUARK pipelines using predicted contact maps in CASP12 , 2018, Proteins.

[32]  Jinbo Xu,et al.  Analysis of distance‐based protein structure prediction by deep learning in CASP13 , 2019, Proteins.

[33]  Mohammed AlQuraishi.  End-to-end differentiable learning of protein structure , 2018, bioRxiv.

[34]  David T. Jones,et al.  Deep learning extends de novo protein modelling coverage of genomes using iteratively predicted structural constraints , 2018, Nature Communications.

[35]  Demis Hassabis,et al.  Improved protein structure prediction using potentials from deep learning , 2020, Nature.