What is Hail's methodology?

Hi - I’m just wondering. I can’t see in the documentation any reference to the methods being employed (i.e. in GWAS e.g. LMM etc) or any citation material if I attempted to publish any findings.

Can someone direct me to it please (apologies if I’ve missed something obvious).


Hail doesn’t have a prepackaged ‘gwas’ method, so the documentation is on each specific method that might be used to do per-variant association tests, which you can find on the statistical methods page:

  • linear_regression_rows
  • logistic_regression_rows
  • poisson_regression_rows
  • linear_mixed_regression_rows

Note that these methods are not called “gwas” because they are useful outside of variant association – these same methods work after aggregating rare variants per gene in a burden test, for instance!

You can absolutely cite the software, there’s a hl.citation method that generates versioned links for reproducibility:

>>> print(hl.citation())
Hail Team. Hail 0.2.13-81ab564db2b4. https://github.com/hail-is/hail/releases/tag/0.2.13.