Pairwise sequence alignment below the twilight zone

Area of Science:

• Bioinformatics
• Structural Biology
• Computational Biology

Background:

• Accurate sequence alignment is crucial for identifying remote protein homologues and for homology modeling.
• Existing methods struggle with low pairwise identity, limiting the discovery of evolutionary relationships.
• Protein substitution patterns differ significantly between closely and distantly related sequences.

Purpose of the Study:

• To develop improved methods for sequence alignment at low pairwise identity.
• To enhance the identification of remote protein homologues and improve fold recognition.
• To leverage structural data for characterizing evolutionary residue-substitution profiles.

Main Methods:

• Development of new amino acid residue interchange matrices derived from structural superposition data.
• Exploitation of known structural homology to characterize evolutionary effects on residue-substitution profiles.
• Application of cyclic permutation strategies to detect homologues with gene duplication and deletion events.

Main Results:

• Structural interchange matrices significantly increase pairwise alignment accuracy for distantly related sequences.
• Demonstrated enhanced functional annotation and fold recognition accuracy using the new matrices.
• Successfully identified novel remote homologues in the Helicobacter pylori genome, including unannotated open reading frames.

Conclusions:

• Structural data provide a superior basis for developing alignment techniques for remote homologues.
• The developed structural interchange matrices offer a chemically sensible and effective approach to improve sequence alignment and protein function prediction.
• The new methods facilitate the discovery of evolutionary relationships and enhance genomic annotation efforts.