What problem does it solve?
The Struct Predictor skill provides accurate and efficient protein structure prediction capabilities, simplifying the analysis and interpretation of molecular structures for users with no advanced training.
Core Features & Use Cases
- Boltz-2 Based: Uses Boltz-2, an efficient tool for biomolecular interactions, for structural predictions.
- Accepts Multi-chain Inputs: Capable of predicting the structure of proteins, especially in cases where proteins have multi-chain complexes.
- Generate Confidence Scores: Calculates and displays confidence scores like pLDDT (Post-Lyophilization Disorder Determination score) and PAE (Predicted Aligned Error) for each residue.
- Generate Reports: Outputs detailed reports in Markdown format, including plots and heatmaps for easy interpretation.
- Demo Mode: Offers a demo mode with pre-defined protein structure (Trp-cage miniprotein) for instant testing without any input file.
- Use Case: A user interested in predicting the structure of a particular protein could use this skill to perform a high-level prediction and analyze the results.
Quick Start
Run the prediction with the provided YAML input file: python skills/struct-predictor/struct_predictor.py --input my_protein.yaml --output /tmp/struct_out
