Strategies of non-sequential protein structure alignments

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Algorithm Analysis

Background:

• Evaluating protein structure alignment algorithms is essential due to numerous available methods.
• Common performance metrics include the number of aligned residues and Root Mean Square Deviation (RMSD).
• Non-sequential structure alignment algorithms often employ varying residue assignment strategies.

Purpose of the Study:

• To analyze the influence of different residue assignment strategies on alignment quality measures (aligned residues and RMSD).
• To investigate how these strategies affect the performance comparison of non-sequential protein structure alignment algorithms.

Main Methods:

• Implemented diverse residue assignment strategies within the GANGSTA+ non-sequential structure alignment algorithm.
• Compared resulting aligned residue counts and RMSD values across strategies and different alignment algorithms.
• Utilized a benchmark set of circular-permuted protein pairs for evaluation.

Main Results:

• Observed significant influence of residue assignment strategies on aligned residue numbers and RMSD.
• Demonstrated that differences in these strategies can strongly bias observed algorithm performance.
• Showed that aligning residue assignment strategies can explain performance variations between algorithms.

Conclusions:

• Residue assignment strategies are a critical, often overlooked, factor in evaluating protein structure alignment algorithms.
• Ignoring these strategies can lead to biased conclusions about algorithm performance.
• Standardizing residue assignment strategies is recommended for reliable and comparable performance assessments of non-sequential protein structure alignment algorithms.