Optimum superimposition of protein structures: ambiguities and implications

Area of Science:

• Structural Biology
• Bioinformatics
• Computational Biology

Background:

• Protein structure comparison and superimposition are vital for analysis, modeling, and classification.
• Structural similarity is often used to infer evolutionary relatedness.
• The uniqueness of protein structure superimposition has been largely overlooked.

Purpose of the Study:

• To investigate the uniqueness of protein structure superimposition.
• To develop and implement an efficient algorithm for finding alternative structural alignments.

Main Methods:

• Developed an algorithm based on optimum superimposition and dynamic programming.
• Tested and validated the algorithm using published data and protein data bank.
• Performed automatic classification of all protein folds in a recent protein data bank release.

Main Results:

• The algorithm identified alternative, equally optimal alignments for many protein pairs.
• These alternative alignments are indistinguishable by standard metrics (equivalent residues, root mean square error).
• Distinct sets of equivalent residue pairs characterize these alternative alignments across all protein architectures.

Conclusions:

• Protein fold superimposition is frequently ambiguous, yielding multiple valid alignments.
• This ambiguity challenges the interpretation of structural similarity for evolutionary inference.
• Studies assuming unique superimposition may lead to misleading conclusions.