Abstract:Residue contact maps provide a condensed two-dimensional representation of three-dimensional protein structures, serving as a foundational framework in structural modeling but also as an effective tool in their own right in identifying inter-helical binding sites and drawing insights about protein function. Treating contact maps primarily as an intermediate step for 3D structure prediction, contact prediction methods have limited themselves exclusively to sequential features. Now that AlphaFold2 predicts 3D structures with good accuracy in general, we examine (1) how well predicted 3D structures can be directly used for deciding residue contacts, and (2) whether features from 3D structures can be leveraged to further improve residue contact prediction. With a well-known benchmark dataset, we tested predicting inter-helical residue contact based on AlphaFold2's predicted structures, which gave an 83% average precision, already outperforming a sequential features-based state-of-the-art model. We then developed a procedure to extract features from atomic structure in the neighborhood of a residue pair, hypothesizing that these features will be useful in determining if the residue pair is in contact, provided the structure is decently accurate, such as predicted by AlphaFold2. Training on features generated from experimentally determined structures, we leveraged knowledge from known structures to significantly improve residue contact prediction, when testing using the same set of features but derived using AlphaFold2 structures. Our results demonstrate a remarkable improvement over AlphaFold2, achieving over 91.9% average precision for a held-out subset and over 89.5% average precision in cross-validation experiments.

Improving AlphaFold 3 structural modeling by incorporating explicit crosslinks

Modelling protein complexes with crosslinking mass spectrometry and deep learning

Modeling flexible protein structure with AlphaFold2 and cross-linking mass spectrometry

Protein structure prediction with in-cell photo-crosslinking mass spectrometry and deep learning

Accurate structure prediction of biomolecular interactions with AlphaFold 3

Cross-linking mass spectrometry discovers, evaluates, and corroborates structures and protein–protein interactions in the human cell

Improved AlphaFold modeling with implicit experimental information

Increasing the Depth of Mass-Spectrometry-Based Structural Analysis of Protein Complexes Through the Use of Multiple Cross-Linkers.

Leveraging Cross-Linking Mass Spectrometry for Modeling Antibody–Antigen Complexes

Unmasking AlphaFold to integrate experiments and predictions in multimeric complexes

Unmasking AlphaFold: integration of experiments and predictions in multimeric complexes

Improving AlphaFold Predicted Contacts for Alpha-Helical Transmembrane Proteins Using Structural Features

Distance-AF: Modifying Predicted Protein Structure Models by Alphafold2 with User-Specified Distance Constraints

AlphaFold3 versus experimental structures: assessment of the accuracy in ligand-bound G protein-coupled receptors

Dissecting AlphaFolds Capabilities with Limited Sequence Information

Utilizing anomalous signals for element identification in macromolecular crystallography

From sequence to mechanobiology? Promises and challenges for AlphaFold 3

Comparative analysis of RNA 3D structure prediction methods: towards enhanced modeling of RNA–ligand interactions

Protein complexes in cells by AI‐assisted structural proteomics

Highly significant improvement of protein sequence alignments with AlphaFold2

Modeling conformational states of proteins with AlphaFold