Analyzing the sequence-structure relationship of a library of local structural prototypes

Area of Science:

• Structural bioinformatics
• Computational biology
• Biophysics

Background:

• Understanding the relationship between amino acid sequences and local 3D protein structures is fundamental in molecular biology.
• Existing methods often struggle with the complexity and variability of local protein folds.

Purpose of the Study:

• To develop and apply a novel computational approach for analyzing sequence-structure relationships in proteins.
• To identify specific amino acid patterns and structural motifs that govern local protein conformation.

Main Methods:

• Utilized an unsupervised clustering approach, the hybrid protein model (HPM), to build a library of local structural prototypes.
• Employed multiple structural alignment of protein folds from a non-redundant databank, encoded into a structural alphabet of 16 protein blocks (PBs).
• Incorporated variable-length fragments and considered gaps in structural prototypes, yielding 120 prototypes.

Main Results:

• Identified 120 local structure prototypes, with 25% containing gaps, allowing for variable fragment lengths.
• Discovered that tight turns predominantly originate from three specific PB series within the HPM.
• Uncovered seven amino acid equivalence classes with strong propensities for specific local structures using multivariate analysis.
• Defined 'contrast factors' highlighting sequence-structure specificities, such as Gly/Asn-rich turns and Pro-rich coils versus Ser/Thr/Asn/Glu-enriched structures.

Conclusions:

• The HPM provides a robust framework for analyzing complex sequence-structure relationships in proteins.
• The identified amino acid equivalence classes and contrast factors offer insights into local protein folding determinants.
• These findings can enhance fragment-based methods for improving protein structure prediction accuracy.