rosettafold-all-atom

name	rosettafold-all-atom
description	Run RoseTTAFold All-Atom for proteins, nucleic acids, small molecules, metals, covalent modifications, and higher-order assemblies. Use when its Hydra input model and confidence metrics fit the problem and licensed dependencies are available.
license	MIT

RoseTTAFold All-Atom

Use $cx-modeling-problem-execution for concrete structures and continue through input preparation, inference, analysis, and review.

Gate and workflow

1. Ask once before cloning a pinned commit, creating the environment/container, downloading large sequence/template databases and weights, installing separately licensed SignalP, and running GPU compute. Verify each dependency's terms.
2. Validate protein/nucleic-acid FASTA, ligand SDF/SMILES, metals, covalent bond atom indices/chirality, chain IDs, and Hydra config. Pin checkpoints, database snapshots, cycles, seeds, hardware, and container digest.
3. Smoke-test an official example. Run into artifacts/<run-id>/rfaa/; retain inputs/config, MSAs/templates, PDB, pLDDT/PAE/PDE metrics, logs, mappings, environment, and failures.
4. Inspect covalent/metal geometry, chirality, ligand state, interfaces, clashes, and confidence; do not reduce quality to one mean metric.
5. Record with $science-provenance; review with $science-review.

Boundaries

• SignalP and other dependencies may have separate licenses and cannot be silently downloaded or redistributed.
• Predicted docks and pae_inter are prioritization signals, not affinity.