BeEP

BeEP assesses protein structural models by leveraging evolutionary substitution patterns to evaluate how well structural constraints inferred from models are represented within sequence alignments of homologous proteins.

Key Features:

• Evolutionary substitution pattern analysis: Uses substitution patterns that arise from structural constraints in homologous proteins sharing a specific conformation.
• Site-specific substitution matrices: Constructs site-specific substitution matrices based on a protein evolution model that incorporates structural data.
• Maximum likelihood evaluation: Evaluates the effectiveness of the substitution matrices using maximum likelihood calculations.
• Position-specific and global scoring: Produces position-specific and global scores reflecting the fit between model-inferred constraints and homologous sequence alignments.
• Model ranking and selection: Ranks and selects protein models, identifying native-like structures.
• No explicit parameterization: Does not rely on explicit parameterization to identify structural similarities between models and target structures.
• CASP validation: Validated on a subset of proteins from the Critical Assessment of techniques for protein Structure Prediction (CASP).

Scientific Applications:

• Model ranking and selection: Ranking protein structural models and selecting native-like candidates based on evolutionary compatibility.
• Model discrimination: Discriminating between alternative structural models using evolutionary-derived scores.
• Assessment of structural constraint representation: Assessing how well structural constraints inferred from models are represented within sequence alignments of homologous proteins.
• Exploration of conformational ensembles: Aiding exploration of the conformational ensemble associated with the native state through evolutionary signals.
• Benchmarking and validation: Benchmarking and validation using data from CASP.

Methodology:

Constructs site-specific substitution matrices from a protein evolution model incorporating structural data; evaluates these matrices using maximum likelihood calculations to produce position-specific and global scores that indicate the representation of model-inferred structural constraints within sequence alignments of homologous proteins, and uses those scores to rank models.