eMatchSite

eMatchSite performs sequence order-independent structure alignments of ligand binding pockets to detect similar binding sites across proteins for drug-target identification and proteome-scale analyses.

Key Features:

• Sequence Order-Independent Alignments: Constructs alignments of ligand binding pockets without relying on residue sequence order, enabling detection of functional similarities despite lack of global homology.
• High Tolerance for Structural Distortions: Maintains alignment accuracy for structurally distorted or low-quality predicted protein models, supporting proteome-scale analyses of gene products.
• Benchmarking Performance: Large-scale benchmarking on adenine-binding pockets showed eMatchSite aligns nearly three times more protein pairs than SOIPPA and incurs only a 4–9% reduction in accuracy compared with crystal structures for weakly homologous models.
• Superiority Over Existing Algorithms: Outperforms SiteEngine by 6% on high-quality models and 13% on moderate-quality models in recognizing similar binding sites.

Scientific Applications:

• Drug–protein interaction network analysis: Enables systematic investigation of drug–protein interaction networks across complete proteomes.
• Polypharmacology: Supports systems-level polypharmacology studies by identifying shared or off-target binding sites.
• Rational drug repositioning: Facilitates rational drug repositioning by revealing binding-site similarities between unrelated proteins.

Methodology:

Performs sequence order-independent alignment of ligand binding pockets and tolerates structural distortions to align predicted protein structure models without requiring high-quality experimental (crystal) structures; benchmarking used adenine-binding pockets.