Gaps in structurally similar proteins: towards improvement of multiple sequence alignment

Area of Science:

• Bioinformatics
• Structural Biology
• Computational Biology

Background:

• Sequence alignments are crucial for understanding protein structure and function.
• Accurate identification and placement of gaps (insertions/deletions) in sequence alignments remain challenging.
• Existing methods for gap optimization often lack sufficient accuracy, impacting downstream analyses.

Purpose of the Study:

• To develop an algorithm for locally optimizing gaps within protein sequence alignments.
• To statistically analyze large datasets of gaps and flanking regions to derive predictive residue preferences.
• To improve the accuracy of sequence alignment algorithms and indel prediction.

Main Methods:

• Developed an algorithm to optimize gaps using the FSSP database, analyzing over 2 million gaps.
• Extracted datasets of non-identical gaps and flanking regions for statistical analysis.
• Derived log-odds scores based on residue frequencies at different positions relative to gaps.

Main Results:

• Gaps were enriched in small, turn-prone residues (D, G, N, P, S) and depleted in hydrophobic residues.
• Flanking regions showed enrichment in hydrophobic residues and higher secondary structure propensity.
• The derived scores accurately predicted gap-containing regions and improved alignment accuracy when integrated into Clustal-W.

Conclusions:

• Statistically derived residue preferences provide significant predictive information for gap placement.
• The developed method enhances the accuracy of pairwise and multiple sequence alignments.
• Applications include improved gap penalties, indel prediction, and automated structure-based alignment refinement.