Request for correction: DiffDock manifest draft in OpenSciFlow

[kkkniengLiu]

Hi DiffDock maintainers,

I am maintaining an early open initiative called OpenSciFlow: https://github.com/OpenSciFlow

We are drafting small, reviewable plugin manifests for AI-for-science tools so local/HPC workflows can record inputs, environment, model weights, commands, outputs, citations, and limitations before execution.

We created a draft DiffDock manifest here:

• Manifest: https://github.com/OpenSciFlow/plugin-manifest/blob/main/examples/diffdock-docking/opensciflow.yaml
• Review notes: https://github.com/OpenSciFlow/plugin-manifest/blob/main/docs/diffdock-review-notes.md
• Landscape entry: https://github.com/OpenSciFlow/awesome-ai4s-workflows/blob/main/categories/chemistry-materials.md

This is not a request for support, adoption, or integration work. I am opening this only to ask whether the draft is accurate enough and whether anything is misleading.

Specific corrections that would help:

1. Should DiffDock and DiffDock-L be represented as one manifest or separate manifests?
2. Is the inference command shape in the manifest reasonable for single-complex and CSV use?
3. What is the best way to represent model-weight retrieval, licenses, and checksums?
4. Which output files should a wrapper treat as stable artifacts?
5. Are the safety boundaries around confidence scores, binding affinity, and small-molecule docking stated correctly?
6. Is there a better canonical citation or docs link we should use?

If GitHub issues are not the right place for this, please feel free to close it. A short correction comment is enough; I can make the change on the OpenSciFlow side.