CoDiFold

CoDiFold integrates contact and distance profile information into an enhanced Rosetta low-resolution energy function to improve de novo protein structure prediction from amino acid sequences.

Key Features:

• Integration of contacts and distance profiles: Combines contact information and distance profiles within the Rosetta low-resolution energy function to refine energy evaluation for predicted conformations.
• Improved energy–RMSD correlation: Enhances the correlation between calculated energies and root mean square deviation (RMSD) to improve selection of native-like models.
• Population-based multi-mutation strategy: Employs a population-based multi-mutation approach for conformation sampling to explore conformational space efficiently.
• Benchmark performance: Achieved 49.24% and 45.21% reductions in average RMSD on a test set of 43 proteins compared to Rosetta and QUARK, respectively.

Scientific Applications:

• De novo protein structure prediction: Predicts tertiary structures from amino acid sequence without relying on homologous templates.
• Modeling proteins with unknown structures: Generates models for proteins lacking experimentally determined structures to support studies of function and interactions.
• CASP benchmarking: Produced predictions comparable to state-of-the-art methods for the 10 free modeling (FM) targets evaluated in CASP13.

Methodology:

Combines contact information and distance profiles within an enhanced Rosetta low-resolution energy function and applies a population-based multi-mutation conformation sampling strategy.