CoChat

High precision protein functional site detection using 3D convolutional neural networks

Open access
Torng, Wen, Altman, Russ B · 2019 · Bioinformatics

The DOI matches its Crossref record; 53 of the 53 references it cites were found in a registry or an index.

Abstract

Motivation: Accurate annotation of protein functions is fundamental for understanding molecular and cellular physiology. Data-driven methods hold promise for systematically deriving rules underlying the relationship between protein structure and function. However, the choice of protein structural representation is critical. Pre-defined biochemical features emphasize certain aspects of protein properties while ignoring others, and therefore may fail to capture critical information in complex protein sites. Results: In this paper, we present a general framework that applies 3D convolutional neural networks (3DCNNs) to structure-based protein functional site detection. The framework can extract task-dependent features automatically from the raw atom distributions. We benchmarked our method against other methods and demonstrate better or comparable performance for site detection. Our deep 3DCNNs achieved an average recall of 0.955 at a precision threshold of 0.99 on PROSITE families, detected 98.89 and 92.88% of nitric oxide synthase and TRYPSIN-like enzyme sites in Catalytic Site Atlas, and showed good performance on challenging cases where sequence motifs are absent but a function is known to exist. Finally, we inspected the individual contributions of each atom to the classification decisions and show that our models successfully recapitulate known 3D features within protein functional sites. Availability and implementation: The 3DCNN models described in this paper are available at https://simtk.org/projects/fscnn. Supplementary information: Supplementary data are available at Bioinformatics online.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1093/bioinformatics/bty813

Cites

53 of 53 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1. Local box extraction. (a) For each recorded functional atom coordinate (dotted sphere), the amino acid which this atom belongs to is identified (highlighted in red) and assigned as the central amino acid. (b) Backbone atoms of the selected amino acid are used to calculate the orthogonal axes for box extraction. (c) Using the defined orientation, a local box of 20 Å is extracted around the selected amino acid, centering on the Cβ atom
Fig. 1. Local box extraction. (a) For each recorded functional atom coordinate (dotted sphere), the amino acid which this atom belongs to is identified (highlighted in red) and assigned as the central amino acid. (b) Backbone atoms of the selected amino acid are used to calculate the orthogonal axes for box extraction. (c) Using the defined orientation, a local box of 20 Å is extracted around the selected amino acid, centering on the Cβ atomCC BY-NC · PubMed Central
Fig. 2. Local box featurization. (a) Structure in each local 20 Å box is decomposed into Carbon, Oxygen, Nitrogen, and Sulfur channels. (b) Each atom type channel structure is further divided into 3D 1-Å voxels, within which the presence of atom of the corresponding atom type is recorded. Gaussian filters are applied to the discrete counts within each channel. (c) The resulting numerical 3D matrices of the four atom types are then stacked together as different input channels
Fig. 2. Local box featurization. (a) Structure in each local 20 Å box is decomposed into Carbon, Oxygen, Nitrogen, and Sulfur channels. (b) Each atom type channel structure is further divided into 3D 1-Å voxels, within which the presence of atom of the corresponding atom type is recorded. Gaussian filters are applied to the discrete counts within each channel. (c) The resulting numerical 3D matrices of the four atom types are then stacked together as different input channelsCC BY-NC · PubMed Central
Fig. 3. The FEATURE program. FEATURE characterizes a specified location in protein structure by dividing the local environment into six concentric shells, each of 1.25 Å in thickness. Within each shell, 80 different physicochemical properties are evaluated, resulting in a numeric vector of length 480
Fig. 3. The FEATURE program. FEATURE characterizes a specified location in protein structure by dividing the local environment into six concentric shells, each of 1.25 Å in thickness. Within each shell, 80 different physicochemical properties are evaluated, resulting in a numeric vector of length 480CC BY-NC · PubMed Central
Fig. 4. Schematic diagram of the models. (a) 3DCNNs take in voxel representation of local protein boxes. The 3D convolutional and max-pooling layers extract 3D biochemical features at different spatial scales and the Softmax classifier calculates the class probabilities. Prediction error drives parameter updates in the Softmax layer and convolutional filters to achieve optimal performance. (b) The Voxel-SVM models take in flattened voxel representation and calculate class probabilities via the SVM classifier. (c) The FEATURE-SVM models use FEATURE vectors as input, and calculate class probabilities via the SVM classifier. In contrast to 3DCNNs, the input features are fixed during training. (d) The FEATURE-1DCNNs extract 1D features from FEATURE vectors and use the Softmax classifier to make predictions
Fig. 4. Schematic diagram of the models. (a) 3DCNNs take in voxel representation of local protein boxes. The 3D convolutional and max-pooling layers extract 3D biochemical features at different spatial scales and the Softmax classifier calculates the class probabilities. Prediction error drives parameter updates in the Softmax layer and convolutional filters to achieve optimal performance. (b) The Voxel-SVM models take in flattened voxel representation and calculate class probabilities via the SVM classifier. (c) The FEATURE-SVM models use FEATURE vectors as input, and calculate class probabilities via the SVM classifier. In contrast to 3DCNNs, the input features are fixed during training. (d) The FEATURE-1DCNNs extract 1D features from FEATURE vectors and use the Softmax classifier to make predictionsCC BY-NC · PubMed Central
Fig. 5. Importance visualization of local functional site microenvironments. The atom importance map displays the importance score (0–100) of each input atom in heat map colors, from the most important (red) to the least important (blue). (a) Microenvironment surrounding a conserved ASP residue in the EF_HAND motif (PDB: 1CKK, ASP 20). (b) Microenvironment surrounding a ASP residue in a PROSITE false negative EF_HAND site (PDB: 2LE9, ASP 64). (c) Microenvironment surrounding a conserved CYS residue in the EGF_1 motif (PDB: 1BF9, CYS 81). (d) Microenvironment surrounding a conserved LYS residue in the RNASE_PANCREATIC motif (PDB: 11BG, LYS 41). (e) Microenvironment surrounding a TRYPSIN_SER motif, centered on a conserved SER residue (PDB: 1ELD, SER 203). (f) Microenvironment surrounding an INSULIN motif, centered on a conserved CYS residue (PDB: 1IOG, CYS 7-chain A)
Fig. 5. Importance visualization of local functional site microenvironments. The atom importance map displays the importance score (0–100) of each input atom in heat map colors, from the most important (red) to the least important (blue). (a) Microenvironment surrounding a conserved ASP residue in the EF_HAND motif (PDB: 1CKK, ASP 20). (b) Microenvironment surrounding a ASP residue in a PROSITE false negative EF_HAND site (PDB: 2LE9, ASP 64). (c) Microenvironment surrounding a conserved CYS residue in the EGF_1 motif (PDB: 1BF9, CYS 81). (d) Microenvironment surrounding a conserved LYS residue in the RNASE_PANCREATIC motif (PDB: 11BG, LYS 41). (e) Microenvironment surrounding a TRYPSIN_SER motif, centered on a conserved SER residue (PDB: 1ELD, SER 203). (f) Microenvironment surrounding an INSULIN motif, centered on a conserved CYS residue (PDB: 1IOG, CYS 7-chain A)CC BY-NC · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

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

Show all 53 references

Cited by

See all 69

Metadata

Authors
Torng, W., & Altman, R. B.
Affiliations
Stanford University (US)
Corresponding author
Altman, Russ B
Venue
Bioinformatics
Published
2018-09-20
Volume
35
Issue
9
Pages
1503-1512
ISSN
  • 1367-4803 (ISSN-L)
  • 1367-4811
OpenAlex
W2891927397
Cited by
69
References
53
Citations by year
2026: 4 · 2025: 6 · 2024: 16 · 2023: 13 · 2022: 8 · 2021: 16 · 2020: 3 · 2019: 3
FWCI
1.96
Type
journal article
Language
en
Open access
Locations
  • Bioinformatics (publishedVersion, cc-by-nc)
  • PubMed Central (publishedVersion, cc-by-nc)
  • PubMed (publishedVersion)
Funders
National Institutes of Health (GM102365, LM05652, HL117798)
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
3D convolutional neural networks, nitric oxide synthase, trypsin-like enzymes, feature attribution, protein structure-function relationship
MeSH terms
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

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: High precision protein functional site detection using 3D convolutional neural networks. https://app.cochat.ai/papers/10.1093/bioinformatics/bty813
The paper
Torng, W., & Altman, R. B. (2019). High precision protein functional site detection using 3D convolutional neural networks. Bioinformatics. https://doi.org/10.1093/bioinformatics/bty813

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.