manifold-genetics

An end-to-end pipeline from PLINK genotypes to publication figures: PCA → admixture → manifold embedding → visualisation → metrics.

Installation

Requires Python 3.10–3.12, on Linux or macOS.

pip install manifold-genetics

Admixture is an optional extra:

pip install 'manifold-genetics[admixture]'

Installation from source

pip install git+https://github.com/MattScicluna/manifold_genetics

Usage

manifold-genetics acquire synthetic     # a simulated cohort and its config, in seconds
manifold-genetics run config.yaml

acquire hgdp fetches the public HGDP+1KGP cohort instead; acquire custom writes a config for PLINK files you already have. For raw biobank genotypes, preprocess filters and intersects SNPs and subsample chooses the fit samples — by label counts, a list, or a label-free geometric sketch — each writing a directory run accepts.

As a Nextflow workflow

nextflow run nextflow/main.nf -profile no_admixture --fit_plink data/fit --labels labels.csv --colormap colormap.json

See Nextflow. The same pipeline is also a set of WDL workflows, including the All of Us ones; see Workflows.

Documentation and tutorials

• Documentation
• Quickstart — a full run on public data, and the input and output formats
• Tutorial — a complete run on a simulated cohort, under a minute, nothing to download
• Install — optional extras, external tools, and developing on the repository
• Preprocessing — filtering raw biobank genotypes and projecting them onto a reference panel

Help

Open an issue.

License

BSD 3-Clause License (see LICENSE).

Citation

If you use this package in your research, please cite:

[Your citation here]