aipoch/open-science / resources/skills/diffdock
diffdock
VerifiedPredict small-molecule binding poses with DiffDock-L (Corso et al. 2023/2024, github.com/gcorso/DiffDock) — blind diffusion docking that places a ligand into a protein pocket without a predefined search box and ranks the samples with a learned confidence model. Reach for this skill to dock a SMILES or SDF against a PDB, to generate ranked 3D poses for a small fragment library, or to get a starting pose for downstream rescoring. DiffDock predicts geometry, not affinity.
| File | Kind | Size |
|---|---|---|
| .catalog_stamp | other | 13 B |
| SKILL.md | skill md | 4.8 KB |
| references/workflows.md | markdown | 1.9 KB |
references/workflows.md
View on GitHub# DiffDock — batch and sequence-only workflows
The main `SKILL.md` covers the single-complex path. Two further entry points
share the same `inference` module and the same YAML-overwrites-CLI caveat.
## Batch / small-library docking
Prepare one CSV row per complex and hand it to `--protein_ligand_csv`:
```csv
complex_name,protein_path,ligand_description,protein_sequence
6lu7_frag001,6lu7.pdb,CC(=O)Nc1ccc(O)cc1,
6lu7_frag002,6lu7.pdb,COc1ccc(C#N)cc1,
```
```bash
cd $DIFFDOCK_REPO
python3 -m inference \
--config default_inference_args.yaml \
--protein_ligand_csv batch.csv \
--out_dir out/
```
Each row gets its own subdirectory under `--out_dir`. `complex_name` may be
left empty (an index is used). `ligand_description` accepts a SMILES string or
a path to an `.sdf`/`.mol2`; one ligand per row, so a multi-ligand SDF needs to
be split first.
For libraries beyond a few hundred ligands, raise `samples_per_complex` only
in a copied YAML (see the CLI-overwrite gotcha) and shard the CSV across jobs —
the embedding step holds ESM-2 650M in GPU memory for the whole run.
## Sequence-only receptor (ESMFold first)
If a receptor structure is not available, leave `protein_path` empty and fill
`protein_sequence`; DiffDock folds the chain with ESMFold before docking. That
path needs `fair-esm` installed; the `esm2_t33_650M_UR50D` checkpoint is
fetched on first use into `~/.cache/torch/hub/checkpoints`. Expect the fold to
dominate wall-clock for receptors over ~400
residues, and treat poses against an ESMFold model with more scepticism than
against a crystal structure: pocket geometry inherits the fold's pLDDT.
## Reading confidence across a library
Confidence logits are calibrated _within_ a complex, not across ligands. To
compare ligands, rescore each `rank1.sdf` with an external scorer; using the
raw DiffDock confidence as a virtual-screening ranker conflates "well-placed
pose" with "tight binder" and the upstream FAQ explicitly warns against it.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 468k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 319k | Updated | Claude CodeCodex | Verified | Install |