binding-affinity
Plan DGL-LifeSci protein-ligand binding affinity workflows with PDBBind, local complexes, ACNN, PotentialNet, complex graph construction, and safe input checks.
- License 许可
- Apache-2.0
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- binding-affinity
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Plan DGL-LifeSci protein-ligand binding affinity workflows with PDBBind, local complexes, ACNN, PotentialNet, complex graph construction, and safe input checks.
- disable-model-invocation
- true
02 资源文件 3
- 参考文档references/data-formats.md6.3 KB
- 参考文档references/troubleshooting.md8.1 KB
- 参考文档references/workflows.md10.0 KB
03 SKILL.md 正文
Binding Affinity
Use this sub-skill when a task involves protein-ligand binding affinity prediction with DGL-LifeSci structure-based workflows.
Use This For
- Planning PDBBind
coreorrefinedworkflows without immediately downloading or training on the full dataset. - Preparing local protein-ligand complexes with ligand files, protein/pocket PDB files, 3D coordinates, and
load_moleculeflags. - Choosing between
ACNNandPotentialNetand matching each model to the graph object it expects. - Constructing complex graphs with
ACNN_graph_construction_and_featurizationorPN_graph_construction_and_featurization. - Running safe, local smoke checks on protein/ligand file paths before launching graph construction or training.
Start Here
- Read
references/data-formats.mdto validate protein/ligand file formats, PDBBind-style layout, coordinate expectations, andload_moleculeflags. - Run
scripts/check_complex_inputs.py --help, then check the user's local complex pair before constructing graphs. - Read
references/workflows.mdto pick ACNN or PotentialNet, build the PDBBind/local-complex plan, and map graph outputs into model inputs. - Use
references/troubleshooting.mdwhen imports, RDKit parsing, missing conformations, PDBBind paths, config choices, or resource limits fail.
Quick Commands
Check only file presence and supported extensions:
python scripts/check_complex_inputs.py --protein pocket.pdb --ligand ligand.sdf
Inspect ligand metadata and require a usable 3D conformation for binding graph construction:
python scripts/check_complex_inputs.py \
--protein pocket.pdb --ligand ligand.sdf --inspect-ligand --remove-hs
Routing Boundaries
Stay in this sub-skill for PDBBind, local protein-ligand complexes, ACNN/PotentialNet binding models, complex graph construction, binding-specific splits, and safe input checks.
Route elsewhere when the task asks for:
- General SMILES/CSV featurizers, MoleculeNet datasets, splitters, or molecular graph validation: use
../molecule-data-prep/SKILL.md. - General model-zoo architecture catalogs, pretrained GIN checkpoints, molecule embeddings, or non-binding model constructors: use
../model-zoo-pretraining/SKILL.md. - Generic molecular property prediction loops, metrics, masking, or CSV train/eval workflows: use
../property-prediction/SKILL.md.
Bundled Resources
references/workflows.md: PDBBind/local complex workflow planning, ACNN/PotentialNet selection, graph construction contracts, and training-smoke guidance.references/data-formats.md: supported protein/ligand file formats, coordinates, binding pockets, PDBBind local paths, and configuration flags.references/troubleshooting.md: import/install, optional dependency, data/config, CLI/API misuse, conformation, RDKit, layout, and resource-limit fixes.scripts/check_complex_inputs.py: offline helper for protein/ligand existence and extension checks, with optional ligandload_moleculemetadata inspection.
Safety Notes
The original binding-affinity example runner is long-running and can trigger PDBBind downloads, multiprocessing molecule loading, GPU/CPU training, and multi-epoch evaluation. Treat it as reference-only; adapt the distilled contracts in this sub-skill and run tiny local checks before any full PDBBind execution.
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.