DNABindProt

DNABINDPROT predicts DNA-binding residues in proteins by analyzing residue fluctuation patterns with the Gaussian network model (GNM) to identify residues and residue pairs indicative of DNA binding.

Key Features:

• Gaussian Network Model (GNM): Uses GNM to analyze residue fluctuations in protein structures.
• High-frequency mode analysis: Focuses on high-frequency modes that are indicative of potential DNA-binding activity.
• Mean-square distance fluctuation detection: Identifies residue pairs exhibiting significant mean-square distance fluctuations.
• Evolutionary conservation filtering: Filters fluctuating residues using evolutionary conservation profiles with a conservation threshold reported as 5.
• Ranking of binding likelihood: Ranks filtered residues according to their likelihood to participate in DNA binding.
• Neighboring-residue consideration: Includes neighboring residue pairs in the analysis to improve sensitivity and precision.
• Benchmarking on unbound-bound structures: Validated on a dataset of 36 unbound-bound protein structure pairs with reported metrics at conservation threshold 5: sensitivity 9.3%, specificity 90.5%, precision 18.1%, accuracy 78.6%; when neighboring residues are included: sensitivity 24.3%, specificity 93.4%, precision 45.3%, accuracy 83.3%.
• Focus on binding core and coupled residues: Emphasizes detection of binding core residues and identifies residues whose fluctuations are coupled with DNA-binding sites, which may also participate in ligand or protein–protein binding.

Scientific Applications:

• DNA-binding site identification: Identification of DNA-binding residues within protein structures for structural analysis.
• Structural biology and bioinformatics research: Application in studies that characterize protein–DNA interactions and binding-site localization.
• Analysis of coupled functional interactions: Detection of residues coupled to DNA-binding sites that may indicate ligand or protein–protein interaction sites.

Methodology:

Analyze residue fluctuations using the Gaussian network model (GNM), focus on high-frequency modes, identify residue pairs with significant mean-square distance fluctuations, filter by evolutionary conservation profiles (threshold 5), rank residues by likelihood, and optionally include neighboring residue pairs in the analysis.