InterPred

InterPred predicts and models protein–protein interactions (PPIs) from protein sequences to produce molecular-detail structural models of protein complexes.

Key Features:

• Sequence-based PPI prediction: Predicts and models PPIs directly from protein sequences.
• Structural modeling: Integrates structural modeling to generate structural representations of interacting proteins and complexes.
• Extensive structural comparisons: Performs extensive structural comparisons to identify potential interaction templates and structural similarities.
• Molecular docking: Applies molecular docking to generate and refine interaction models of protein pairs.
• Random forest classifier: Implements a novel random forest classifier that incorporates multiple structural features to distinguish correct from incorrect PPI models.
• Full-atom complex modeling: Produces full-atom protein–protein complex models and has demonstrated superior performance to state-of-the-art docking methods on benchmark sets.
• Validation against experimental data: Performance validated against experimental high-throughput techniques with comparable or superior detection efficacy.
• CAPRI37 performance: Ranked among the top predictors in the CAPRI37 experiment.

Scientific Applications:

• PPI detection: Detection of protein–protein interactions with performance comparable or superior to experimental high-throughput techniques.
• Structural characterization: Generation of full-atom models of protein–protein complexes for molecular-level analysis.
• Benchmarking and method evaluation: Comparison and evaluation of docking and interaction-prediction methods using standard benchmark sets and CAPRI experiments.

Methodology:

InterPred integrates structural modeling, extensive structural comparisons and molecular docking, and scores generated interaction models using a random forest classifier that incorporates multiple structural features.