pipeline¶

Multi-stage study orchestration added in v2.6.0.

Public API¶

from openpkflow.pipeline import (
    PipelineConfig,
    StudyPipeline,
    StudyPipelineResult,
    load_pipeline_config,
)

Symbol	Role
`PipelineConfig`	Paths and method options for optional stages
`load_pipeline_config(path)`	Load JSON (YAML if PyYAML installed); paths relative to config file
`StudyPipeline(config).run()`	Run enabled stages; raise on failure
`StudyPipelineResult`	Aggregated results + audit `metadata`
`.summary()`	ASCII multi-section text + disclaimer
`.report(path)`	HTML (default) or Markdown by extension
`.audit_bundle(path)`	ZIP with copied inputs, config, results, HTML report, and SHA-256 manifest
`.to_dict()`	JSON-serializable audit payload

Stages¶

Stage	Enabled when	Library used
Dissolution	`dissolution_csv` + reference/test labels	`DissolutionStudy`
NCA	`nca_csv`	`NCAStudy` (explicit `nca_auc_method`, `nca_blq_method`)
BE	`be_csv`	`BEStudy` TOST

IVIVC is not wired yet (needs multi-array inputs, not a single CSV).

API and web app¶

The FastAPI adapter exposes:

POST /api/pipeline/analyze
POST /api/pipeline/report
POST /api/pipeline/audit-bundle

The React page at /pipeline accepts one to three stage uploads, requires explicit NCA AUC and BLQ methods when NCA is enabled, renders unified results, and downloads the report or audit bundle. All calculations remain in src/openpkflow/.

CLI¶

openpkflow study run examples/study_pipeline_example.json --report out.html
openpkflow study run config.json --json out.json

Examples¶

examples/study_pipeline_example.json
examples/study_pipeline_example.yaml (needs PyYAML)
examples/pipeline_walkthrough.py (sequential API demo without pipeline package)

Tutorial¶

See Formulation-to-report pipeline.