genal

Polygenic risk scoring (PRS) and Mendelian randomization (MR) in Python.

Quick links

• Project docs: https://genal.readthedocs.io

• Paper (citation): https://doi.org/10.1093/bioadv/vbae207

• Source code: https://github.com/CypRiv/genal

Documentation

• Introduction
  • Design notes
  • When to use genal (and when not)
• Setup
  • Install
  • PLINK 2
  • Configuration and cache locations
  • Reference data (two different concepts)
  • Offline usage
• Core concepts
  • The Geno object model
  • Standard columns
  • Method behaviors (what you get back)
  • Side effects and files
  • Resource usage (ram, cpus)
• Tutorial (detailed)
  • 0) Mental model: what a Geno instance represents
  • 1) Create a Geno from GWAS summary statistics
  • 2) Preprocess and standardize the table
  • 3) Select independent instruments via LD clumping
  • 4) Compute a PRS in a target cohort (individual-level genotypes)
  • 5) Two-sample MR: query an outcome GWAS and run MR
  • 6a) MR-PRESSO (outlier detection and distortion testing)
  • 6b) Leave-one-out MR (sensitivity analysis)
  • 7) Additional capabilities (beyond the core pipeline)
• Methods (implementation notes)
  • Per-variant F-statistic (FSTAT)
  • MR harmonization
  • Two-sample MR estimators
  • Leave-one-out MR
  • MR-PRESSO (summary)
  • Colocalization (ABF)
• API reference
  • Top-level helpers (import genal)
  • genal.Geno
• FAQ / Troubleshooting
  • PLINK path is not set
  • preprocess_data() can’t find rsIDs / many SNPs are missing
  • PRS uses far fewer SNPs than expected
  • Clumping returns no SNPs
  • MR results look wrong for binary outcomes
  • Palindromic SNPs are removed/flipped
  • GWAS Catalog queries fail or time out
  • tmp_GENAL/ is large
  • PLINK “out of memory”

Citation

If you use genal, please cite:

Cyprien A. Rivier, Santiago Clocchiatti-Tuozzo, Shufan Huo, Victor Torres-Lopez, Daniela Renedo, Kevin N. Sheth, Guido J. Falcone, Julian N. Acosta.
Genal: A Python Toolkit for Genetic Risk Scoring and Mendelian Randomization.
Bioinformatics Advances (2024). DOI: https://doi.org/10.1093/bioadv/vbae207

Additional method citations

If you use the following tools/methods through genal, please also cite the original method/tool papers:

• MR-PRESSO (via genal.Geno.MRpresso()): Marie Verbanck, Chia‑Yen Chen, Benjamin Neale, Ron Do. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nature Genetics (2018). DOI: https://doi.org/10.1038/s41588-018-0099-7. PMID: 29686387.

• Colocalization (ABF) (via genal.Geno.colocalize()): Claudia Giambartolomei et al. Bayesian test for colocalisation between pairs of genetic association studies using summary statistics. PLoS Genetics (2014). DOI: https://doi.org/10.1371/journal.pgen.1004383.

• PLINK 2 (used for clumping/PRS/association tests/allele frequencies): Chang C.C. et al. Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience (2015). DOI: https://doi.org/10.1186/s13742-015-0047-8.