How pinder-org/pinder works
PINDER provides protein–protein docking data, tooling to generate and inspect that resource, and evaluation and prediction workflows. The principal user path is to obtain the published dataset, load indexed structures through pinder-core, and evaluate docking results. The graph also preserves data-ingestion, QC/plotting, and method-runner capabilities evident from the README and tree. Wiring for several unsampled modules is intentionally left unspecified; edges reflect only documented behavior or sampled calls.
Main entry points: utils.py Index and paths · data_pipeline.py Ingestion pipeline · main.py Method actions · system.py Pinder system
Components 26
Dataset access
- Index and paths
src/pinder-core/pinder/core/index/utils.py - Pinder system
src/pinder-core/pinder/core/index/system.py - Dataset loader
src/pinder-core/pinder/core/loader/loader.py - Structure operations
src/pinder-core/pinder/core/loader/structure.py - Local dataset
Data production
- Ingestion pipeline
src/pinder-data/pinder/data/pipeline/data_pipeline.py - Task batching
src/pinder-data/pinder/data/pipeline/tasks.py - Dimer construction
src/pinder-data/pinder/data/get_dimers.py - Annotations
src/pinder-data/pinder/data/annotation - Apo pairing
src/pinder-data/pinder/data/get_apo.py - Dataset splits
src/pinder-data/pinder/data/get_splits.py - Similarity clusters
src/pinder-data/pinder/data/get_clusters.py
Quality and analysis
- Dataset QC
src/pinder-data/pinder/data/qc - Dataset plots
src/pinder-data/pinder/data/plot
Docking evaluation
- Evaluation runner
src/pinder-eval/pinder/eval/run.py - Docking metrics
src/pinder-eval/pinder/eval/dockq/method.py - Submission creation
src/pinder-eval/pinder/eval/create_submission.py
Prediction methods
- Method actions
src/pinder-methods/pinder/methods/main.py - Inference preparation
src/pinder-methods/pinder/methods/preprocess - FRODOCK predictor
src/pinder-methods/pinder/methods/run/frodock/predict.py - HDOCK predictor
src/pinder-methods/pinder/methods/run/hdock/predict.py - PatchDock predictor
src/pinder-methods/pinder/methods/run/patchdock/predict.py
Other components
- Researcher
- Published dataset
- Cloud storage
- RCSB structures
Connections 30
- Researcher → Index and paths: locates dataset
- Published dataset → Index and paths: serves index
- Index and paths → Cloud storage: downloads files
- Index and paths → Local dataset: syncs dataset
- Researcher → Dataset loader: requests data
- Dataset loader → Index and paths: gets index
- Dataset loader → Local dataset: loads files
- Researcher → Pinder system: selects complex
- Pinder system → Index and paths: resolves entry
- Pinder system → Structure operations: aligns structures
- Ingestion pipeline → Task batching: builds task batches
- Ingestion pipeline → Dimer construction: runs ingestion tasks
- Ingestion pipeline → Annotations: runs annotations
- Ingestion pipeline → Apo pairing: runs apo pairing
- Ingestion pipeline → Dataset splits: generates splits
- Ingestion pipeline → Similarity clusters: generates clusters
- Task batching → RCSB structures: discovers structures
- Dimer construction → Local dataset: writes dataset files
- Researcher → Dataset QC: checks dataset
- Dataset QC → Index and paths: reads index
- Researcher → Dataset plots: analyzes results
- Researcher → Evaluation runner: evaluates predictions
- Evaluation runner → Docking metrics: computes metrics
- Evaluation runner → Index and paths: reads evaluation index
- Researcher → Submission creation: creates submission
- Researcher → Method actions: chooses action
- Method actions → Inference preparation: prepares inputs
- Method actions → FRODOCK predictor: dispatches prediction
- Method actions → HDOCK predictor: dispatches prediction
- Method actions → PatchDock predictor: dispatches prediction