HVIDB

HVIDB catalogs experimentally validated and predicted human–virus protein–protein interactions and associated annotations to support analysis of host–virus molecular mechanisms.

Key Features:

• Extensive Data Repository: Contains 48,643 experimentally validated human–virus PPIs spanning 35 virus families, including 6,633 virally targeted host complexes, 3,572 host dependency/restriction factors, and 911 verified or predicted three-dimensional (3D) complex structures of human–virus PPIs.
• Gene Expression Profiles: Provides tissue-specific expression profiles for 6,790 human genes targeted by viral interactions and includes data from 129 Gene Expression Omnibus (GEO) series reporting differentially expressed genes following viral infections.
• Computational Prediction Tools: Integrates machine learning models that use sequence embedding, interolog mapping, and domain–domain interaction inference to predict potential human–virus PPIs.

Scientific Applications:

• Mechanistic studies: Enables analysis of molecular mechanisms underlying viral infection by linking PPIs, host complexes, and host dependency/restriction factors.
• Hypothesis-driven experiments: Supports prioritization of candidate host or viral proteins for experimental validation based on curated PPIs and predictive scores.
• Antiviral target discovery: Informs identification of host or viral targets for therapeutic intervention through integrated PPI, structural, and expression data.
• Structural analysis: Facilitates examination of verified or predicted 3D human–virus PPI complexes to study interaction interfaces.

Methodology:

HVIDB integrates experimentally verified PPIs and applies machine learning models incorporating sequence embedding, interolog mapping, and domain–domain interaction inference to predict human–virus PPIs.