SCORE: predicting the core of protein models

Area of Science:

• Computational biology
• Structural bioinformatics
• Protein structure prediction

Background:

• Accurate homology model generation relies on identifying reliable regions derived from template structures.
• Current methods for defining these regions have limitations, often depending on family size and secondary structure definitions.

Purpose of the Study:

• To develop and evaluate a novel algorithm, SCORE, for identifying and building the core regions of homology models.
• To improve the accuracy and efficiency of homology model generation.

Main Methods:

• The SCORE algorithm utilizes phi, psi constraints to identify conserved regions across protein families.
• It employs environmentally constrained substitution tables to extend these conserved regions.
• The algorithm rapidly identifies and builds homology model cores from sequence alignments.

Main Results:

• SCORE identifies structurally related regions from basis structures, moving beyond conventional core definitions.
• The algorithm achieves rapid identification and model core construction, typically under 1 second.
• Evaluation on 114 model cores showed high accuracy, with cores comprising over 50% of the structure in most cases and RMSD of 3.7 Å or less.

Conclusions:

• The SCORE algorithm provides a fast and accurate method for identifying and building homology model cores.
• Its approach using novel constraints enhances the reliability of the most accurate parts of homology models.
• SCORE offers a significant improvement over existing methods for homology modeling.