ThreaDom

ThreaDom predicts protein domain boundaries by threading sequences against the Protein Data Bank (PDB) and computing residue-level conservation scores to identify continuous and discontinuous domains.

Key Features:

• Template-Based Approach: Threads a target protein sequence through the PDB library to identify templates with similar structural folds.
• Domain Conservation Score (DCS): Computes a per-residue DCS that integrates information from template domain structures, terminal and internal alignment gaps, and insertions.
• Multiple Threading Alignments: Combines multiple threading alignments to enhance prediction of both continuous and discontinuous domains consisting of separated sequence segments.

Scientific Applications:

• Single and Multi-Domain Classification: On 630 non-redundant sequences, correctly classified single versus multi-domain proteins in 81% of cases and assigned domain linkers within ±20 residues in 78% of instances.
• Discontinuous Domain Prediction: On 486 proteins with discontinuous domains, achieved an average precision of 84% and a recall rate of 65%.
• CASP Evaluations: On CASP8 targets reported domain overlap rates of 73% for Free Modeling, 87% for hard multiple-domain proteins, and 85% for discontinuous-domain proteins, with consistent results in CASP9 and CASP10 evaluations.

Methodology:

Thread the target sequence through the PDB library; compute per-residue DCS from template domain structures, alignment gaps and insertions; combine multiple threading alignments and derive domain boundaries from the distribution of DCS profiles.