Identification of ligand-binding residues using protein sequence profile alignment and query-specific support vector...

Area of Science:

• Computational biology
• Bioinformatics
• Protein science

Background:

• Understanding protein functions relies on information within ligand-binding residues (LBRs).
• Accurately identifying LBRs solely from protein sequences remains a significant challenge in bioinformatics.

Purpose of the Study:

• To develop a novel computational method, I-LBR, for identifying LBRs using only protein sequence data.
• To offer both general-purpose (I-LBRGP) and ligand-specific (I-LBR LS) prediction modes.

Main Methods:

• I-LBR employs a query-specific approach, constructing training subsets based on input sequence information.
• The Support Vector Machine (SVM) algorithm is utilized to train the LBR identification model.
• Residue probabilities for LBR classification are predicted for query proteins.

Main Results:

• The ligand-specific mode (I-LBR LS) demonstrates superior performance when the query protein's ligand type is known.
• I-LBR achieves performance comparable or superior to existing state-of-the-art LBR identification methods.
• Experimental validation was conducted on four distinct testing datasets.

Conclusions:

• I-LBR provides an effective sequence-based computational strategy for identifying ligand-binding residues.
• The ligand-specific mode offers enhanced accuracy for targeted LBR prediction.
• The I-LBR web server and associated datasets are publicly available for research purposes.