ZoomQA

ZoomQA assesses residue-level accuracy of tertiary protein structure predictions for single proteins and protein complexes to identify local modeling errors and inform structural analysis.

Key Features:

• Residue-Level Accuracy Assessment: Provides residue-level estimates of local model accuracy for tertiary protein structures of single proteins and protein complexes.
• Fragment-Based Feature Analysis: Evaluates changes in chemical and physical properties within a fragment surrounding a target amino acid within a specified radius r.
• Fourteen Physicochemical Residue Properties: Represents each residue using fourteen distinct physical and chemical amino-acid properties to build local structural descriptors.
• Performance Benchmarking: Benchmarked against other local QA methods and CASP14 global prediction metrics, demonstrating superior performance in identifying inaccurate model regions.
• Homology-Independent Assessment: Performs assessments without homology searching or Position-Specific Scoring Matrices (PSSM).

Scientific Applications:

• Drug Discovery: Informs drug discovery by identifying residue-level inaccuracies that affect ligand-binding sites and interaction interfaces.
• Structural Model Refinement: Guides refinement of predicted protein models by pinpointing problematic regions at the residue level.
• Viral Protein Analysis: Identifies problematic regions within complex proteins such as SARS-CoV-2 proteins to aid structural biology investigations.

Methodology:

Constructs fragment-based residue representations within a radius r using fourteen physical and chemical amino-acid properties, computes residue-level accuracy scores, benchmarks against other local QA methods and CASP14 metrics, and performs assessments without homology searches or PSSM.