[Placeholder] - API for Exploratory Statistics

[jerowe]

This is waiting on the user story issue Exploratory Statistics.

Maybe something like this -

sgkit/
   stats (or filter?)/
      variants
      samples

Decisions

How much of the filtering and subsetting do we want to be in sgkit? We cannot cover every case, but should we try to cover what is in the tour or skallele blog post?

This isn't a proposed API, just thinking things on 'paper'.

sgkit.stats.variant.quality()
sgkit.stats.variant.bin('MQ', bin_size) # look at a binned distribution of MQ 
sgkit.stats.sample.missingness()
sgkit.stats.variant.count_segregating()

[hammer]

I think API design discussions are best done in the main sgkit repo. https://github.com/pystatgen/sgkit/issues/37 and https://github.com/pystatgen/sgkit/issues/29 are relevant to this discussion.

We've previously discussed having a layered API in which the base layer tries to stay as close to PyData ecosystem APIs as possible, and then having a higher level API on top of that to expose functionality to less sophisticated users.

[jerowe]

@hammer great! Thanks! I'll take a look at the API discussions.