DroMel positive DFE

[clararehmann]

Addresses #1690

I used the Huber et al. 2017 gamma-distributed DFE with a proportion of nonsynonymous mutations being positively selected. Parameters for the positively-selected mutations were from Table S6 in the Zhen et al. paper.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 99.85%. Comparing base (8a42a8f) to head (a37cf56).
Report is 4 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1695   +/-   ##
=======================================
  Coverage   99.85%   99.85%           
=======================================
  Files         139      139           
  Lines        4774     4827   +53     
  Branches      470      470           
=======================================
+ Hits         4767     4820   +53     
  Misses          3        3           
  Partials        4        4           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[clararehmann]

@petrelharp - I noticed that in this implementation you used the Huber 2017 model w/o singletons, but in the implementation of the HomSap DFE with positive mutations (#1694) you used the full model. Is this on purpose?

[petrelharp]

Gee - I don't remember this at all, and haven't been able to reconstruct what this means - could you bump my memory? (What "singletons"?)

[petrelharp]

Oh wait I understand. Hm let's see.

[petrelharp]

For reference, table S2 from H17:

[petrelharp]

And, here's the table from Zhan et al showing which parameters they actually used:

As you note, Zhen et al (according to this table) are using the "all data" estimates for both species, but for some reason I've taken the "no singletons" numbers for DroMel. Good catch!

I think the parameters for DroMel should be

    gamma_shape = 0.35  # shape
    gamma_mean = -1.33e-04  # expected value

I'm getting the "mean" values from the H17 table, rather than taking beta from the Z24 table and multiplying. Please check! (I also got worried I mixed up 'mean' and 'beta' for HomSap, but it looks okay!) If you agree, could you swap those in, for both the implementation and QC?