Area-Based Selection of Binding Interfaces for Structural Prediction of Protein-Protein Complexes
Yang, Y. X.; Zhu, B. T.
Show abstract
Protein-protein interaction is a fundamental process in all biological systems, and the structural information of a protein-protein complex may provide important mechanistic details and insights into the biological processes involved. Elucidation of the rules underlying the interface specificity in protein-protein interactions is of great value for the correct prediction of the structures of a protein-protein complex. In the present study, we have developed the area-based methods for selecting near-native interfaces for protein-protein interactions. The quantitative relationship between different areas in the predicted structure of protein-protein complex and the predicted accuracy was explored using linear and nonlinear models. The predicted accuracy is characterized using the root mean square deviation (L_RMSD) of ligands. The performances of the newly-developed area-based models for selecting near-native interfaces for protein-protein binding interactions based on the partners structures at unbound or bound states are better than (or at least comparable to) those of the existing, more sophisticated method(s). The success rates of some models are above 90% (some are close to 100%), which indicates the importance and effectiveness of the area-based interface selection. The area-based methods presented in this work may shed lights on the final resolution of the interface selection problem in the field of protein-protein complex structure prediction and also on the rules of interface specificity for protein-protein interactions from a geometric area perspective. The principles developed this work also shed lights on understanding the protein-protein binding mechanisms from an area perspective. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=158 HEIGHT=200 SRC="FIGDIR/small/680849v1_ufig1.gif" ALT="Figure 1"> View larger version (61K): org.highwire.dtl.DTLVardef@fc8db3org.highwire.dtl.DTLVardef@bb0f62org.highwire.dtl.DTLVardef@ef1b84org.highwire.dtl.DTLVardef@1f79d8b_HPS_FORMAT_FIGEXP M_FIG C_FIG
Matching journals
The top 7 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Deep learning based predictive modeling to screen natural compounds against TNF-alpha for the potential management of Rheumatoid Arthritis: Virtual screening to comprehensive in silico investigation 96%
- Molecular dynamics simulations reveal the selectivity mechanism of structurally similar agonists to TLR7 and TLR8 96%
- Prediction of Essential Binding Domains for the Endocannabinoid N-Arachidonoylethanolamine (AEA) in the Brain Cannabinoid CB1 receptor 96%
Similar papers in this journal
Similar papers in this journal
- Identification of Family-Specific Features in Cas9 and Cas12 Proteins: A Machine Learning Approach Using Complete Protein Feature Spectrum 95%
- Are Deep Learning Structural Models Sufficiently Accurate for Virtual Screening? Application of Docking Algorithms to AlphaFold2 Predicted Structures 95%
- The R346K Mutation in the Mu Variant of SARS-CoV-2 Alter the Interactions with Monoclonal Antibodies from Class 2: A Free Energy of Perturbation Study 95%
Similar papers in this journal
- Assessment of AI-based Protein Structure Prediction for the NLRP3 Target 96%
- DeepBindGCN: Integrating Molecular Vector Representation with Graph Convolutional Neural Networks for Accurate Protein-Ligand Interaction Prediction 96%
- The Network Basis for the Structural Thermostability and the Functional Thermoactivity of Aldolase B 96%
Similar papers in this journal
- An insight into SARS-CoV-2 Membrane protein interaction with Spike, Envelope, and Nucleocapsid proteins 96%
- A strategy to optimize the peptide-based inhibitors against different mutants of the spike protein of SARS-CoV-2 96%
- Investigating the role of N-terminal domain in phosphodiesterase 4B-inhibition by molecular dynamics simulation 96%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.