DaReUS-Loop

DaReUS-Loop models loop regions in protein homology models to improve structural accuracy of variable and long loop conformations.

Key Features:

• Loop modeling and remodeling: Models and remodels loop regions in protein structures derived from homology models.
• Parallel loop handling: Simultaneously models up to 20 loop regions in parallel.
• Data-driven candidate identification: Mines the complete Protein Data Bank (PDB) to identify potential loop candidates.
• Local conformation profile–profile comparison: Uses local conformation profile–profile comparison to select candidate loop conformations.
• Physico-chemical scoring: Applies physico-chemical scoring to filter and rank candidate loops.
• Prediction confidence score: Assigns a prediction confidence score for each loop model that correlates with expected accuracy.
• Improved long-loop accuracy: Demonstrates improved accuracy for long loops and increases the number of accurately modeled loops on template-based test sets.
• Cross-family sourcing: Reports that over 50% of successful loop models are derived from unrelated proteins, indicating structural convergence beyond sequence homology.

Scientific Applications:

• Structural biology: Improving the accuracy of loop regions in protein homology models used in structural biology studies.
• Modeling variable loops: Modeling and remodeling long and sequence-variable loop regions that deviate from homologous templates.
• Benchmarking: Evaluating loop modeling performance on template-based test sets.
• Structural convergence studies: Investigating instances where similar loop conformations arise across unrelated proteins.

Methodology:

Data-driven mining of the complete PDB, local conformation profile–profile comparison, physico-chemical scoring to filter candidates, and parallel modeling of up to 20 loop regions.