PoSSuM

PoSSuM performs large-scale similarity searches of ligand-binding sites within protein structures to identify and catalog similar binding pockets in the Protein Data Bank (PDB) for protein function prediction and ligand discovery.

Key Features:

• Comprehensive Database: A relational database cataloging known and putative binding sites from the Protein Data Bank (PDB), yielding over 14 million similar binding-site pairs and expanding coverage beyond SitesBase (~33,000 known sites).
• Novel Algorithmic Approach: Uses the multiple-sketches algorithm to perform exhaustive all-pairs similarity searches among millions of binding sites, enabling detection of similar pockets across proteins with different global folds.
• Functional Prediction and Drug Discovery: Facilitates inference of potential ligands for unbound or unannotated structures and supports identification of target proteins based on binding-site similarity for ligand screening.

Scientific Applications:

• Protein Function Prediction: Infers functional relationships for proteins with unknown annotations by identifying similar ligand-binding sites across structurally diverse proteins.
• Drug Discovery and Ligand Identification: Aids in proposing potential ligands and screening target proteins by mapping binding-site similarity relationships derived from PDB entries.

Methodology:

Exhaustive all-pairs similarity searches using the multiple-sketches algorithm across binding sites cataloged in the PDB, with results organized in a relational database.