Significant improvement in accuracy of multiple protein sequence alignments by iterative refinement as assessed by...

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Biology

Background:

• Multiple sequence alignment (MSA) is fundamental for analyzing protein families and inferring evolutionary relationships.
• Accurate MSAs are essential for various downstream applications, including structure prediction and functional annotation.
• Existing alignment methods vary in performance, necessitating evaluation and improvement.

Purpose of the Study:

• To comparatively assess the performance of four distinct strategies for multiple sequence alignment.
• To identify superior alignment methods, particularly for challenging datasets with low sequence identity.
• To evaluate the impact of different strategies on alignment accuracy relative to structural alignments.

Main Methods:

• Evaluated four MSA methods against structural alignments from 54 protein families.
• Optimized gap penalties for each family and method.
• Compared pairwise, progressive, and iterative alignment strategies, including a novel doubly nested iterative approach.

Main Results:

• All four multiple alignment methods outperformed the conventional pairwise method in accuracy.
• A novel iterative strategy optimizing a weighted sum-of-pairs score achieved the highest accuracy.
• Alignment accuracy improvement with iterative methods increased as average sequence identity decreased, highlighting their efficacy for remote homology detection.

Conclusions:

• Iterative refinement strategies, particularly those optimizing weighted sum-of-pairs scores, offer superior performance for multiple sequence alignment.
• The newly developed doubly nested iterative method demonstrates significant potential for aligning distantly related protein sequences.
• The choice of substitution matrix has a minor impact on the relative performance of MSA methods.