VisGrid

VisGrid characterizes local geometric features of protein tertiary structures to annotate surface elements and identify ligand-binding pockets for structure-based functional inference.

Key Features:

• Visibility Criterion: Computes the fraction of visible directions from a target position on a protein surface to quantify local geometry.
• Comprehensive Surface Analysis: Characterizes large protrusions, hollows, and flat regions in addition to potential ligand-binding sites.
• Pockets Identification: Identifies 95.0% of ligand-binding sites as one of the three largest pockets in a benchmark set of 5616 proteins.
• Negative Image Analysis: Detects large protrusions by analyzing the negative image of protein structures.
• Robustness to Structural Flexibility: Evaluated on molecular dynamics–generated distorted structures, showing ~20% sensitivity decrease at RMSD = 2.0 Å while specificity remains largely unaffected.

Scientific Applications:

• Structure-based protein annotation: Infers functional properties from geometric features of protein surfaces.
• Annotation of proteins of unknown function: Provides surface-based descriptors that aid functional inference for uncharacterized proteins.
• Ligand-binding site discovery: Ranks and identifies candidate pockets to support identification of potential ligand-binding sites for drug discovery.

Methodology:

Calculates a visibility fraction from target surface positions (visibility criterion), analyzes negative-image representations of structures, ranks pockets to detect ligand-binding sites within the three largest pockets on a benchmark of 5616 proteins, and evaluates performance on molecular dynamics–derived distorted structures reporting sensitivity and specificity changes at specified RMSD values.