Identification of homologous core structures.

Area of Science:

• Structural bioinformatics
• Protein evolution
• Computational biology

Background:

• Distinguishing homologous (evolutionarily related) and analogous (convergent evolution) protein structures is challenging, especially when sequence similarity is undetectable.
• Traditional measures like root-mean-square (RMS) deviation and percentage of identical residues fail to reliably differentiate remote homologs from analogs.

Purpose of the Study:

• To develop and validate a novel method for accurately distinguishing homologous from analogous protein structural neighbors.
• To identify a conserved substructure indicative of homology that is absent in analogous structures.

Main Methods:

• Utilized a large database of protein structure-structure alignments.
• Developed and tested a method based on identifying a conserved substructure, termed the homologous core structure (HCS).
• Employed cross-validation to assess the performance of the HCS test against traditional pairwise similarity measures.

Main Results:

• Pairwise structural similarity measures (RMS, identical residues) poorly distinguish remote homologs from analogs.
• Analogous structures rarely superimpose the specific substructure shared by homologous proteins.
• The HCS test achieved 75% accuracy in identifying remote homologs with a 16% false-positive rate for analogs.

Conclusions:

• The homologous core structure (HCS) provides a robust method for identifying remote protein homologs.
• HCS serves as a structural analog to sequence motifs, enabling the detection of evolutionary relationships at greater distances.
• This method significantly improves upon existing techniques for classifying protein structural relationships.