journal article·2022-07Open access

Encoding protein dynamic information in graph representation for functional residue identification

Yuan Chiang, Wei-Han Hui, Shu-Wei Chang

Cell Reports Physical Science· Volume 3· Issue 7

Abstract

Recent advances in protein function prediction exploit graph-based deep learning approaches to correlate the structural and topological features of proteins with their molecular functions. However, proteins in vivo are not static but dynamic molecules that alter conformation for functional purposes. Here, we apply normal mode analysis to native protein conformations and augment protein graphs by connecting edges between dynamically correlated residue pairs. In the multilabel function classification task, our method demonstrates a remarkable performance gain based on this dynamics-informed representation. The proposed graph neural network, ProDAR, increases the interpretability and generalizability of residue-level annotations and robustly reflects structural nuance in proteins. We elucidate the importance of dynamic information in graph representation by comparing class activation maps for hMTH1, nitrophorin, and SARS-CoV-2 receptor binding domain. Our model successfully learns the dynamic fingerprints of proteins and pinpoints the residues of functional impacts, with vast untapped potential for broad biotechnology and pharmaceutical applications.
Source: openalex_inverted_index · Quality: baseline

Research context

OpenAlex topics: Protein Structure and Dynamics · Computational Drug Discovery Methods · Machine Learning in Bioinformatics

SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.

References (70)

Show source-provided references
1. Drenth · 2007
2. How cryo-em is revolutionizing structural biology10.1016/j.tibs.2014.10.005 ↗
3. Cryo-electron microscopy methodology: current aspects and future directions10.1016/j.tibs.2019.04.008 ↗
4. Extraction of protein dynamics information from cryo-em maps using deep learning10.1038/s42256-020-00290-y ↗
5. Announcing the worldwide protein data bank10.1038/nsb1203-980 ↗
6. Uniprot: the universal protein knowledgebase in 202110.1093/nar/gkaa1100 ↗
7. Comparative protein structure modeling using modeller10.1002/cpbi.3 ↗
8. Swiss-model: homology modelling of protein structures and complexes10.1093/nar/gky427 ↗
9. Highly accurate protein structure prediction with alphafold10.1038/s41586-021-03819-2 ↗
10. Accurate prediction of protein structures and interactions using a three-track neural network10.1126/science.abj8754 ↗
11. Se (3)-transformers: 3d rototranslation equivariant attention networks
12. Generative models for graphbased protein design
13. Normal mode analysis of biomolecular structures: functional mechanisms of membrane proteins10.1021/cr900095e ↗
14. Structural and kinetic studies of the human nudix hydrolase mth1 reveal the mechanism for its broad substrate specificity10.1074/jbc.M116.749713 ↗
15. Towards region-specific propagation of protein functions10.1093/bioinformatics/bty834 ↗
16. Cath functional families predict functional sites in proteins10.1093/bioinformatics/btaa937 ↗
17. Structure-based protein function prediction using graph convolutional networks10.1038/s41467-021-23303-9 ↗
18. Unsupervised protein embeddings outperform hand-crafted sequence and structure features at predicting molecular function10.1093/bioinformatics/btaa701 ↗
19. Proteingcn: protein model quality assessment using graph convolutional networks
20. Persgnn: applying topological data analysis and geometric deep learning to structure-based protein function prediction
21. Structural dynamics is a determinant of the functional significance of missense variants10.1073/pnas.1715896115 ↗
22. Ensemble-based computational approach discriminates functional activity of p53 cancer and rescue mutants10.1371/journal.pcbi.1002238 ↗
23. Exploring allosteric pathways of a v-type enzyme with dynamical perturbation networks10.1021/acs.jpcb.9b01294 ↗
24. Pdbe: towards reusable data delivery infrastructure at protein data bank in europe10.1093/nar/gkx1070 ↗
25. Rcsb protein data bank: powerful new tools for exploring 3d structures of biological macromolecules for basic and applied research and education in fundamental biology, biomedicine, biotechnology, bioengineering and energy sciences10.1093/nar/gkaa1038 ↗
26. Protein data bank Japan (pdbj): updated user interfaces, resource description framework, analysis tools for large structures
27. New tools and functions in data-out activities at protein data bank Japan (pdbj)10.1002/pro.3273 ↗
28. Sifts: structure integration with function, taxonomy and sequences resource10.1093/nar/gks1258 ↗
29. Sifts: updated structure integration with function, taxonomy and sequences resource allows 40-fold increase in coverage of structure-based annotations for proteins10.1093/nar/gky1114 ↗
30. Gradcam: visual explanations from deep networks via gradient-based localization
31. Persistent homology-a survey10.1090/conm/453/08802 ↗
32. Persistence images: a stable vector representation of persistent homology
33. Semi-supervised classification with graph convolutional networks
34. Inductive representation learning on large graphs
35. Protein function prediction via graph kernels10.1093/bioinformatics/bti1007 ↗
36. How powerful are graph neural networks?
37. High precision protein functional site detection using 3d convolutional neural networks10.1093/bioinformatics/bty813 ↗
38. Predicting protein function from domain content10.1093/bioinformatics/btn312 ↗
39. Learning deep features for discriminative localization
40. Role of binding site loops in controlling nitric oxide release: structure and kinetics of mutant forms of nitrophorin 410.1021/bi049748a ↗
41. Structure, function, and antigenicity of the sars-cov-2 spike glycoprotein10.1016/j.cell.2020.11.032 ↗
42. Structure of the sars-cov-2 spike receptor-binding domain bound to the ace2 receptor10.1038/s41586-020-2180-5 ↗
43. Broad host range of sars-cov-2 and the molecular basis for sars-cov-2 binding to cat ace210.1038/s41421-020-00210-9 ↗
44. Cryo-em structures of mers-cov and sars-cov spike glycoproteins reveal the dynamic receptor binding domains10.1038/ncomms15092 ↗
45. Generalized correlation for biomolecular dynamics10.1002/prot.20784 ↗
46. Estimating mutual information
47. Generalized correlation-based dynamical network analysis: a new high-performance approach for identifying allosteric communications in molecular dynamics trajectories10.1063/5.0018980 ↗
48. Torsional network model: normal modes in torsion angle space better correlate with conformation changes in proteins10.1103/PhysRevLett.104.228103 ↗
49. Why are large conformational changes well described by harmonic normal modes?10.1016/j.bpj.2021.10.027 ↗
50. A new set of amino acid descriptors and its application in peptide qsars10.1002/bip.20296 ↗
51. Where did the blosum62 alignment score matrix come from?10.1038/nbt0804-1035 ↗
52. Learning protein sequence embeddings using information from structure
53. Amino acid encoding for deep learning applications10.1186/s12859-020-03546-x ↗
54. Large amplitude elastic motions in proteins from a single-parameter, atomic analysis10.1103/PhysRevLett.77.1905 ↗
55. Anisotropy of fluctuation dynamics of proteins with an elastic network model10.1016/S0006-3495(01)76033-X ↗
56. Anisotropic network model: systematic evaluation and a new web interface10.1093/bioinformatics/btl448 ↗
57. Principal component analysis: a method for determining the essential dynamics of proteins10.1007/978-1-62703-658-0_11 ↗
58. Edelsbrunner · 2010
59. Topological descriptors for 3d surface analysis
60. The gudhi library: simplicial complexes and persistent homology
61. 3D alpha shapes
62. Prody: protein dynamics inferred from theory and experiments10.1093/bioinformatics/btr168 ↗
63. Biopython: freely available python tools for computational molecular biology and bioinformatics10.1093/bioinformatics/btp163 ↗
64. Hagberg · 2008
65. Schrödinger
66. Vmd: visual molecular dynamics10.1016/0263-7855(96)00018-5 ↗
67. Adam: a method for stochastic optimization
68. Decoupled weight decay regularization
69. Layer normalization
70. Fast graph representation learning with pytorch geometric