PheWAS software
PublicphewasFlow
An R package for reproducible phenome-wide association studies in both directions, with local and cluster workflows, multiple-testing correction, and Manhattan and volcano plots.
phewasFlow applies a consistent analysis specification across clinical phenotypes and records the outcome of every requested model, including those that were skipped or failed.
Two directions for a PheWAS
| Research question | Analysis direction |
|---|---|
| Is one polygenic score, exposure, or biomarker associated with many outcomes? | Phenotypes as outcomes |
| Is each phenotype associated with one fixed clinical endpoint? | Phenotypes as predictors |
The package supports binary, continuous, count, and ordinal outcomes using Firth logistic, linear, negative-binomial, and proportional-odds regression, respectively.
From inputs to results
Start with a participant table, phenotype metadata, and a YAML analysis configuration. phewasFlow validates these inputs, fits the models, combines results across shards, and applies multiple-testing correction. It fits associations using prepared phenotypes; it does not derive them from raw clinical codes.
Run locally or use the repository’s SLURM templates for larger scans. Results include model status, warnings, sample counts, and effect estimates, alongside Manhattan and volcano plots in PNG and PDF formats.
Get started
The package requires R 4.4 or later. Follow the installation instructions and simulated example to try the workflow before supplying your own data. The repository also includes workflow guides and SLURM examples.
Lars Fritsche maintains phewasFlow under the MIT license. When using it in research, cite the software and report the version used; citation("phewasFlow") provides the reference from R.