LFM-Pro: a tool for detecting significant local structural sites in proteins

Area of Science:

• Structural bioinformatics
• Computational biology
• Protein structure analysis

Background:

• Protein structure repositories enable discovery of functional and evolutionary relationships.
• Identifying conserved, family-specific structural sites is crucial for pinpointing important atoms and residues.
• Existing methods for detecting local structural features are computationally intensive and often miss biologically significant details.

Purpose of the Study:

• To introduce Local Feature Mining in Proteins (LFM-Pro), a novel framework for automated discovery of protein family-specific local sites and their associated features.
• To develop a computationally efficient method that overcomes limitations of current approaches in detecting biologically significant local features.

Main Methods:

• LFM-Pro utilizes the distance field to backbone atoms to identify geometrically significant structural centers within a protein.
• Feature vectors are constructed from the geometric and biochemical environments surrounding these centers.
• A statistical measure scores features for their discriminative power between a target protein family and unrelated proteins, with successful features forming a representative set.

Main Results:

• The LFM-Pro framework was successfully applied to the trypsin-like serine proteases family.
• Performance was validated on a challenging protein classification dataset, showing comparable or superior results to DALI.
• The method demonstrated efficacy in both identifying distinctive sites for a protein family and classifying proteins based on extracted features.

Conclusions:

• LFM-Pro provides an effective and efficient approach for discovering family-specific local structural sites and features.
• The framework aids in identifying functionally important residues and enhances protein classification capabilities.
• The software and datasets are publicly available for academic research.