Skip to content

Getting Started

Status Applies to Owner
Pre-release draft main branch as of 2026-07-01 public quick orientation Public user docs

Use this page to choose the shortest path through the user guide. The project is still pre-release, so prefer the repository README and these docs over assumptions from older branches or external examples.

If You Are New To g

Start with Installation, then run the command checks from the same checkout and environment that will run your scan:

uv run g --help
uv run g regenie --help

Then use Quickstart for concrete quantitative, binary, approximate-Firth, and GPU command shapes.

If You Have A REGENIE Workflow

The active user workflow is REGENIE Step 2 over BGEN input:

  1. Generate Step 1 prediction lists with upstream regenie.
  2. Run g regenie with quantitative (--qt) or binary (--bt) trait mode.
  3. Inspect the run directory, manifest, Parquet parts, and logs.

Use Compatibility to check whether the workflow is currently supported. g does not implement Step 1, BED/PGEN inputs, SPA, categorical covariates, or exact Firth.

If You Need Exact Contracts

Read the reference page for the contract you are touching:

Need Page
Required input files and sample alignment Input Files
Parquet dataset directories and schema Output Files
Resume behavior and manifests Resume and Manifest
CLI flags and exit behavior CLI
TOML merge order and effective config Configuration
Statistical model and output interpretation Algorithm

If You Are Debugging

First confirm the command surface:

uv run g --help
uv run g regenie --help

Then use Troubleshooting for symptom-specific checks. For GPU jobs, also verify the target node with GPU and Clusters.

Development Setup

If you are changing code, building documentation, or using repository fixture data recipes, use the separate Development Installation section.