PconsFold

PconsFold predicts protein tertiary structures ab initio by combining evolutionary contact predictions from PconsC with the Rosetta folding protocol to generate structural models.

Key Features:

• Enhanced Contact Prediction: PconsC separates direct from indirect evolutionary signals to improve contact prediction accuracy for protein families with sufficient sequence data.
• Benchmark Performance: In a benchmark of 15 proteins PconsFold models showed an average 33% improvement in TM-score relative to EVfold, with broader assessments reporting 15–30% improvements over earlier contact prediction methods.
• Rosetta Folding Protocol: Uses Rosetta for structure generation, which improved the chemical realism of models compared to CNS.
• Modularity: The pipeline permits substitution of alternative contact prediction methods.

Scientific Applications:

• Ab initio structure prediction: Predicts three-dimensional protein structures from amino acid sequences using evolutionary contact information.
• Protein family analysis: Improves structural models for protein families with extensive evolutionary data.
• Functional and mechanistic inference: Provides models to support studies of protein function, interactions, and mechanisms.
• Applied research: Supports applications in drug discovery, enzyme engineering, and analysis of disease-related mutations.

Methodology:

Contact prediction with PconsC distinguishing direct and indirect evolutionary signals; structure generation using the Rosetta folding protocol; benchmarking via TM-score comparisons to EVfold and comparisons of model chemical realism to CNS.