docking-inference
Plan, validate, command-build, and troubleshoot DiffDock command-line docking prediction runs for single complexes or batch CSV inputs.
- License 许可
- MIT
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- docking-inference
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Plan, validate, command-build, and troubleshoot DiffDock command-line docking prediction runs for single complexes or batch CSV inputs.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/cli-reference.md6.1 KB
- 参考文档references/configuration.md5.0 KB
- 参考文档references/input-output-formats.md5.0 KB
- 参考文档references/troubleshooting.md9.1 KB
03 SKILL.md 正文
Docking Inference
Use this sub-skill when a user wants to run, adapt, or debug DiffDock command-line docking prediction with python -m inference. It covers single-complex and batch CSV inputs, model/config choices, output layout, confidence-ranked SDF files, optional reverse-process visualization, and safe validation before expensive runs.
Start Here
- Check runtime prerequisites in the root install/runtime reference when that file is present, especially the Torch, PyG, RDKit, ProDy, ESM/OpenFold, and CUDA notes.
- Pick the input mode in input-output-formats.md: batch CSV or single complex.
- Build a dry command with scripts/build_inference_command.py; it prints a command and never runs inference.
- Validate CSV or single-complex inputs with scripts/validate_inference_inputs.py; it uses only Python standard-library modules.
- Review cli-reference.md, configuration.md, and troubleshooting.md before launching a long model run.
Routes
- Batch docking: use a CSV with
complex_name,protein_path,ligand_description, andprotein_sequence; see input-output-formats.md. - Single-complex docking: provide either
--protein_pathor--protein_sequence, plus--ligand_description; see cli-reference.md. - Model/checkpoint overrides: use
--model_dir,--confidence_model_dir,--ckpt, and--confidence_ckpt; see configuration.md. - Output inspection: expect one output directory per complex and confidence-ranked SDF names; see input-output-formats.md.
- Failure recovery: diagnose imports, CPU/GPU behavior, model downloads, invalid inputs, RDKit parsing, ESMFold sequence paths, and GNINA caveats in troubleshooting.md.
Boundaries
- For the graphical Gradio wrapper and browser UI, route to web-ui.
- For benchmark metrics, aggregate RMSD reporting, and evaluation datasets, route to evaluation-benchmarks.
- For training, dataset preparation, and ESM embedding preparation for evaluation/training, route to training-data.
- This sub-skill does not run DiffDock, download models, import heavy DiffDock modules, or require the original repository examples to remain available.
04 同仓库更多 Skills
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
python-api
Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP/mAP semantics, and source-style import diagnostics.
file-evaluation
Folder-based PASCAL VOC AP/mAP evaluation from ground-truth and detection text files.
object-detection-metrics
Guides legacy Object-Detection-Metrics PASCAL VOC AP/mAP evaluation from detection text folders and direct Python API classes.
nx-workspace-patterns
Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.