Profppikernel

Profppikernel predicts protein-protein interactions by applying profile-kernel support vector machines to evolutionary profiles derived from multiple sequence alignments to capture functional and structural protein characteristics.

Key Features:

• Evolutionary profiles: Uses evolutionary profiles derived from multiple sequence alignments to represent functional and structural information of proteins.
• Profile-kernel SVMs: Implements profile-kernel support vector machines (SVMs) that integrate evolutionary information into the prediction model.
• Low-similarity sensitivity: Enhances prediction accuracy for proteins with limited sequence similarity to known interaction partners.
• Gene expression filtering: Applies gene expression data filtering to prioritize highly reliable protein-protein interactions, increasing precision at the expense of recall.

Scientific Applications:

• Functional annotation: Predicts interactions to support annotation of proteins lacking experimental interaction data.
• Drug discovery: Identifies novel protein-protein interactions that can inform target discovery and validation.
• Systems biology: Contributes to accurate modeling of interaction networks for systems-level analyses.

Methodology:

Computational steps explicitly include deriving evolutionary profiles from multiple sequence alignments, employing profile-kernel SVMs that integrate evolutionary information, and applying gene expression data filtering to predicted PPIs.