Interactome INSIDER

Interactome INSIDER integrates genomic variant information with structural protein-protein interactomes and uses machine learning to predict interaction interfaces and assess variant impacts.

Key Features:

• Coverage: Predicts interfaces across 185,957 protein interactions spanning human and seven model organisms and includes the entire experimentally determined human binary interactome.
• Machine learning-driven interface prediction: Employs machine learning algorithms to identify previously unresolved protein interaction interfaces.
• Functional validation: Validation includes data from 2,164 de novo mutagenesis experiments showing that mutations in predicted and known interface residues disrupt interactions at comparable rates.
• Variant enrichment analysis: Performs enrichment analyses of population-wide variants, disease-associated mutations, and somatic (recurrent) cancer mutations within known and predicted interfaces.

Scientific Applications:

• Molecular basis of disease: Links genomic variations to structural interactomes to aid interpretation of disease-associated and cancer-associated mutations.
• Mutation impact assessment: Maps the potential functional consequences of genetic variants on protein-protein interactions.
• Target prioritization: Supports prioritization of interface residues and variants for downstream experimental or therapeutic investigation.

Methodology:

Uses machine learning to predict protein interaction interfaces and conducts enrichment analyses, with predictions compared to results from 2,164 de novo mutagenesis experiments and analyses of disease-related and recurrent cancer mutation enrichment.