A comparison of scoring functions for protein sequence profile alignment.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Protein sequence alignment methods are crucial for understanding protein function and evolution.
• Profile-profile alignment offers potential improvements over traditional sequence-based methods.
• Limited knowledge exists regarding the comparative performance of various profile-profile scoring functions.

Purpose of the Study:

• To evaluate the alignment accuracy of 23 distinct profile-profile scoring functions.
• To compare the performance of profile-profile alignment against established sequence-sequence and profile-sequence methods.
• To identify optimal scoring functions and profile generation strategies for protein sequence alignment.

Main Methods:

• Utilized 488 protein sequence pairs with <30% identity for evaluation.
• Optimized parameters for all scoring functions using a dedicated training set.
• Employed profiles from PSI-BLAST and SAM-T99, with structural alignments from FSSP and CE as benchmarks.
• Compared results against BLAST and PSI-BLAST.

Main Results:

• Profile-profile alignment demonstrated a 2-3% improvement over profile-sequence alignment and a ~40% improvement over sequence-sequence alignment.
• No significant performance differences were observed among most tested scoring functions.
• Profiles generated using SAM-T99 alignments yielded superior results compared to those from PSI-BLAST.

Conclusions:

• Profile-profile alignment represents a significant advancement in protein sequence analysis accuracy.
• The choice of profile generation method (SAM-T99 vs. PSI-BLAST) impacts alignment performance.
• Further research into scoring function optimization may yield additional benefits.