Closed wleoncio closed 3 years ago
BTW, this is only useful if we're really using greedyMix()
in the end: an alternative would be to import and expand the loading functions from rhierbaps (or fastbaps) and dropping the translated greedyMix altogether.
Jukka makes a compelling case for this:
I would strongly advocate transferring the clustering function from either fastbaps or RhierBAPS to rBAPS for the reason that computations are not performed efficiently enough in greedyMix for large modern datasets. While BAPS development itself stopped around 2013, the world of genetic data has gone forward with a speed.
To fix issue #13, one important step would be to write unit tests to the subfunctions, instead of going top-bottom and writing the first test as the one for
greedyMix()
itself.test-greedyMix.R
to add (empty, place-holding) tests for each subfunction