Improving protein secondary structure prediction with aligned homologous sequences

Area of Science:

• Computational biology
• Bioinformatics
• Protein structure prediction

Background:

• Protein secondary structure prediction methods often leverage multiple sequence alignments (MSAs) to enhance accuracy.
• Understanding the interplay between sequence variability, gaps, and secondary structures is crucial for refining prediction algorithms.

Purpose of the Study:

• To investigate the relationship between secondary structural elements, gaps, and variable residue positions in MSAs.
• To assess how these relationships compare to those in structurally aligned protein families.
• To improve protein secondary structure prediction using these insights.

Main Methods:

• Analysis of secondary structure elements, gaps, and variable residue positions in MSAs.
• Comparison with structurally aligned protein families.
• Application of the Quadratic-Logistic method with profiles for prediction.
• Evaluation of the impact of the number of homologous sequences on prediction quality.

Main Results:

• Helical regions exhibit greater variability than coil regions, contrary to expectations.
• Gaps in MSAs serve as a strong indicator for coil regions, enhancing prediction when coil propensity is increased in these areas.
• Prediction accuracy plateaus after approximately 14 homologous sequences.
• Alignment quality significantly impacts prediction accuracy more than the number of homologues.

Conclusions:

• Incorporating gap information, specifically increasing coil propensity in gap regions, improves secondary structure prediction.
• While helical regions show higher variability, this does not directly translate to improved prediction accuracy.
• High-quality alignment of even a few homologous sequences is more beneficial than using a large number of poorly aligned sequences.