SNfold

SNfold implements a sequential niche multimodal conformation sampling algorithm to improve protein structure prediction by enhancing conformational exploration and mitigating local-minima trapping and energy-model inaccuracies.

Key Features:

• Sequential Niche Technique: Applies a sequential niche approach that uses prior sampling information to guide subsequent sampling rounds.
• Derating Function: Uses a derating function derived from previous sampling to construct modified energy landscapes that discourage re-sampling of explored regions.
• Sampling-Guided Energy Functions: Constructs sampling-guided energy functions that facilitate traversal of high-energy barriers and enable multimodal conformational exploration.
• Improved Sampling Efficiency: Achieves more than 100-fold improvement in sampling efficiency compared with Rosetta restrained by distance (Rosetta-dist).
• Prediction Accuracy: Correctly folds 231 out of 300 benchmark proteins with TM-score ≥ 0.5.
• Comparative Performance: Performs comparably to four state-of-the-art methods on CASP13 FM targets within the CASP13 server group.

Scientific Applications:

• Protein structure prediction: Enhances conformational sampling to produce more native-like three-dimensional protein models with improved TM-scores.
• Benchmarking and method evaluation: Enables comparative assessment against methods such as Rosetta-dist and evaluation on CASP13 FM targets.

Methodology:

Designs a derating function based on previous sampling rounds and constructs sampling-guided energy functions to traverse high-energy barriers and prevent redundant sampling.