SciDEX — Task: [Demo] End-to-end provenance demo: trace reasoning

Create a showcase provenance walkthrough: Paper (PubMed) -> KG edges (extracted) -> Hypothesis (generated) -> Dataset (referenced) -> Analysis (executed) -> Model (built) -> Figure (produced). Display the full DAG with artifact versions pinned at each step. Demonstrate verify_reproducibility() showing all inputs are intact. This is the capstone demo for the artifact versioning system. Depends on: a17-24-REPR0001, d16-23-BMOD0001. ## REOPENED TASK — CRITICAL CONTEXT This task was previously marked 'done' but the audit could not verify the work actually landed on main. The original work may have been: - Lost to an orphan branch / failed push - Only a spec-file edit (no code changes) - Already addressed by other agents in the meantime - Made obsolete by subsequent work **Before doing anything else:** 1. **Re-evaluate the task in light of CURRENT main state.** Read the spec and the relevant files on origin/main NOW. The original task may have been written against a state of the code that no longer exists. 2. **Verify the task still advances SciDEX's aims.** If the system has evolved past the need for this work (different architecture, different priorities), close the task with reason "obsolete: " instead of doing it. 3. **Check if it's already done.** Run `git log --grep=''` and read the related commits. If real work landed, complete the task with `--no-sha-check --summary 'Already done in '`. 4. **Make sure your changes don't regress recent functionality.** Many agents have been working on this codebase. Before committing, run `git log --since='24 hours ago' -- ` to see what changed in your area, and verify you don't undo any of it. 5. **Stay scoped.** Only do what this specific task asks for. Do not refactor, do not "fix" unrelated issues, do not add features that weren't requested. Scope creep at this point is regression risk. If you cannot do this task safely (because it would regress, conflict with current direction, or the requirements no longer apply), escalate via `orchestra escalate` with a clear explanation instead of committing.

Completion Notes

Auto-completed by supervisor after successful deploy to main

Git Commits (3)

[Docs] Mark d16-24-PROV0001 acceptance criteria complete2026-04-13

[Docs] Update d16-24-PROV0001 spec work log: verify_reproducibility implemented, demo page added2026-04-13

[Forge] End-to-end provenance demo: verify_reproducibility, /api/analyses/{id}/provenance, /demo/provenance-chain2026-04-13

Spec File

[Demo] End-to-end provenance demo: trace reasoning chain from paper to model

Goal

The capstone demonstration of the artifact versioning system: show a complete, auditable reasoning chain from a published paper through every stage of the SciDEX pipeline to a trained model and its visualizations. Every link in the chain has pinned artifact versions, and the entire chain can be verified for reproducibility.

The Provenance Chain

PubMed Paper (PMID:XXXXX)
  ↓ [extracts] 
KG Edges (gene-disease associations, rate constants)
  ↓ [generates]
Hypothesis (microglial activation in AD)
  ↓ [references]
External Dataset (SEA-AD Allen Brain Cell Atlas)
  ↓ [derives]
Tabular Dataset (differential expression results)
  ↓ [inputs to]
Analysis (gap analysis + model building)
  ↓ [produces]
Biophysical Model (microglial ODE, v1)
  ↓ [generates]
Figures (time-course, sensitivity, fit quality)

Artifact IDs in the Chain

Each node is a registered, versioned artifact:

Paper: paper-{pmid} (v1) — existing PubMed paper about TREM2/microglia

KG Edges: kg_edge-{batch_id} (v1) — edges extracted from that paper

Hypothesis: hypothesis-h-seaad-v4-26ba859b (v1) — existing hypothesis

Dataset: dataset-allen_brain-SEA-AD-MTG-10x (v1) — from d16-21

Tabular Data: tabular_dataset-{id} (v1) — from d16-22

Analysis: analysis-SDA-{id} (v1) — the model-building analysis

Model: model-biophys-microglia-v1 (v1) — from d16-23

Figures: figure-timecourse-{id} (v1), figure-sensitivity-{id} (v1) — from d16-23

Demo Walkthrough

Step 1: Assemble the Chain

Identify or create artifact_links connecting all 8 artifacts:

links = [
    ("kg_edge-batch-001", "paper-12345", "derives_from"),
    ("hypothesis-h-seaad-...", "kg_edge-batch-001", "derives_from"),
    ("tabular_dataset-xxx", "dataset-allen_brain-SEA-AD-MTG-10x", "derives_from"),
    ("analysis-SDA-xxx", "hypothesis-h-seaad-...", "cites"),
    ("analysis-SDA-xxx", "tabular_dataset-xxx", "cites"),
    ("model-biophys-xxx", "analysis-SDA-xxx", "derives_from"),
    ("figure-tc-xxx", "model-biophys-xxx", "derives_from"),
    ("figure-sens-xxx", "model-biophys-xxx", "derives_from"),
]

Step 2: Pin All Versions

Ensure the analysis has pinned_artifacts recording the exact version and content_hash of each input.

Step 3: Verify Reproducibility

result = verify_reproducibility("analysis-SDA-xxx")
assert result["reproducible"] == True
# All pinned inputs exist and content_hash matches

Step 4: Render Provenance DAG

Call GET /api/analysis/{id}/provenance and render the full DAG:

8 nodes (one per artifact type, colored distinctly)
8+ edges showing derivation flow
Version numbers and tags on each node
Green checkmarks for verified artifacts
Interactive: click any node to view artifact detail

Step 5: Showcase Page

Create or extend the demo page to include:

"Provenance Chain" section with the rendered DAG
Narrative walkthrough explaining each link
"Verify Reproducibility" button that runs the check live
Summary: "This model's predictions can be traced back to [paper title] and verified against [dataset name] — every intermediate step is versioned and auditable."

Acceptance Criteria

☑ All 8 artifacts in the chain are registered and linked

☑ artifact_links form a connected DAG from paper to figures

☑ All input artifacts have pinned versions in the analysis

☑ verify_reproducibility() returns True for the chain

☑ Provenance DAG renders with all nodes and edges

☑ Nodes colored by artifact type with version labels

☑ DAG is interactive (clickable nodes)

☑ Demo page includes narrative walkthrough

☑ "Verify Reproducibility" demonstrates live verification

☑ Work log updated with timestamped entry

Dependencies

a17-24-REPR0001 (reproducible analysis chains / verify_reproducibility)
d16-23-BMOD0001 (biophysical model — the chain must exist first)

Work Log

2026-04-14 02:52 PT — Slot minimax:55

Task claimed and evaluated: dependency a17-24-REPR0001 (verify_reproducibility) not on main, only on task branch 99e571466
Decision: implement verify_reproducibility() and provenance API as part of this task to make demo work
Investigated existing data: analysis-SDA-PROV-DEMO-001 already exists with all 8 chain artifacts registered
Implemented verify_reproducibility() in scidex/atlas/artifact_registry.py: checks pinned artifacts exist and content hashes match
Implemented get_analysis_provenance(): combines provenance graph + pinned artifacts + workflow steps + reproducibility
Added GET /api/analyses/{id}/provenance endpoint (returns full DAG + verification)
Added POST /api/analyses/{id}/verify-reproducibility endpoint (runs verification)
Added /demo/provenance-chain page: Cytoscape DAG viz, narrative chain, live Verify button
Tested: verify_reproducibility('analysis-SDA-PROV-DEMO-001') → reproducible=True, 4/4 artifacts verified
Committed and pushed: c8c5b62ca

Payload JSON

{
  "completion_shas": [
    "524da90d0754a2e8bb249a8d935807f866cb43f6",
    "a418f9b55766709979138f5972b2574b9e500049",
    "c8c5b62caf3ee43be0f122c7aef0f630b2807664"
  ],
  "completion_shas_checked_at": "2026-04-14T02:55:24.689630+00:00"
}