Skip to content

Getting started

This page gets you from “EA term list” to “audited decision objects” in minutes.

Install

pip install llm-pathway-curator

Option B: from source

git clone https://github.com/<ORG>/LLM-PathwayCurator.git
cd LLM-PathwayCurator
pip install .

Minimal run (demo)

llm-pathway-curator run \
  --sample-card examples/demo/sample_card.json \
  --evidence-table examples/demo/evidence_table.tsv \
  --out out/demo/

Inspect:

  • out/demo/audit_log.tsv
  • out/demo/report.md

Your first real run

You need two inputs:

  1. EvidenceTable (TSV): term_id, term_name, source, stat, qval, direction, evidence_genes
  2. Sample Card (JSON): structured study context (condition/tissue/perturbation/comparison)

Then:

llm-pathway-curator run \
  --sample-card sample_card.json \
  --evidence-table evidence_table.tsv \
  --out out/run1/

Determinism & provenance

Every run writes run_meta.json. Treat it as the “receipt”:

  • inputs (paths + hashes when available)
  • parameters (τ, k, seed, gates)
  • tool version and backend info

For paper-level reproduction, see paper/README.md and paper/FIGURE_MAP.csv.

Optional: enable LLM proposal mode

LLM-PathwayCurator can use an LLM for proposal-only steps. The audit remains mechanical.

Example (conceptual):

export OPENAI_API_KEY="..."
export LLMPATH_BACKEND="openai"

llm-pathway-curator run \
  --sample-card sample_card.json \
  --evidence-table evidence_table.tsv \
  --out out/llm/

If no API key is provided, the tool should still support a deterministic baseline proposal path.

Build docs locally

From repo root:

pip install -r docs/requirements.txt
mkdocs serve

Open: http://127.0.0.1:8000

Next