Status/maturity of hl.skat

Our team is performing a rare variant analysis, and we’d like to implement a SKAT test. I’ve looked through the forum, and the consensus on hl.skat seems to be that it’s a bit brittle (e.g. Big picture issues: considering switching to HAIL).

Has there been any additional development of hl.skat? What is the current state of this function?

This is a good question, though I’m not sure I have a great answer. The individual who contributed the SKAT implementation was a summer student a couple of years ago, and nobody on the team now is confident in their ownership of that code. It is tested and should work, but might be more prone to things like memory problems than other methods.

That said, we’re working hard on linear algebra infrastructure that will allow us to write SKAT on top of Hail’s Python frontend, rather than as a module in the Scala backend. We’re hoping that this functionality is in place before the end of the year, but it may take a bit longer. Reimplementing SKAT in this way will require us to ensure it is as robust as the rest of the Hail library.

Thanks!