Quat-2L

Quat-2L predicts protein quaternary structural attributes from amino acid sequences using a two-layer classification approach that integrates functional-domain analysis with sequence-correlated pseudo amino acid composition.

Key Features:

• Two-Layer Prediction System: A first layer classifies proteins as monomer, homo-oligomer, or hetero-oligomer and a second layer further classifies predicted oligomers into dimer, trimer, tetramer, pentamer, hexamer, or octamer.
• Hybrid Feature Representation: Integrates functional-domain analysis with sequence-correlated pseudo amino acid composition for input feature construction.
• Performance Metrics: First-layer identification achieves 71.14% overall success; second-layer success rates are 76.91% for homo-oligomers and 82.52% for hetero-oligomers.
• Validation Approach: Performance was validated using jackknife cross-validation on a benchmark dataset curated to ensure no pairwise sequence identity exceeds 60% within subsets.

Scientific Applications:

• Quaternary Structure Prediction: Predicting protein quaternary states from sequence data to inform analyses of protein assembly.
• Structure–Function Studies: Supporting investigation of protein structure–function relationships by providing quaternary state hypotheses.
• Structural Biology: Assisting structural biology efforts by supplying predicted oligomeric states for experimental planning and interpretation.
• Molecular Modeling and Drug Design: Informing molecular modeling and drug design workflows that require knowledge of oligomeric state and assembly.

Methodology:

Uses functional-domain analysis and sequence-correlated pseudo amino acid composition within a two-layer classifier (layer 1: monomer/homo/hetero; layer 2: dimer/trimer/tetramer/pentamer/hexamer/octamer), with performance assessed by jackknife cross-validation on a dataset curated to ≤60% pairwise sequence identity.