compute DE+ and DE- separately

[LevyNat]

No description provided.

[canergen]

Fixed the wrong behavior that we have seen in different models (high significance despite low mean LFC but high standard deviation). It still requires a release note and it would be good to check in our DE tutorial whether things are changing there. I’m not sure that the Bayes factor computation is correct @PierreBoyeau?

[PierreBoyeau]

Hi, LGTM. I am unsure about the Bayes factor here, since we don't have two, but three competing models now(equally expressed, upreg, and downreg). I would be in favor of not computing BFs in this case.

[canergen]

I guess it's fine to remove. I would assume we do two tests: upregulated vs rest and downregulated vs rest. We output only the winner of the test for upregulated/downregulated. I guess the corresponding Bayes factor for the returned FDR is then: upregulated/(equally+downregulated) or downregulated/(equally+upregulated).

[canergen]

Why not return probe_de?

[LevyNat]

updated

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.67%. Comparing base (abfcbfc) to head (15fc4b5).

❗ There is a different number of reports uploaded between BASE (abfcbfc) and HEAD (15fc4b5). Click for more details.

HEAD has 18 uploads less than BASE

Flag	BASE (abfcbfc)	HEAD (15fc4b5)
	21	3

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2826      +/-   ##
==========================================
- Coverage   89.43%   82.67%   -6.77%     
==========================================
  Files         185      185              
  Lines       16182    16182              
==========================================
- Hits        14473    13379    -1094     
- Misses       1709     2803    +1094     

see 27 files with indirect coverage changes

[LevyNat]

@canergen @PierreBoyeau can we merge to main?

I put the new proba_de computation in mode='change_threeway' and kept 'vanilla' as default. Just need to think about how we explain the fix in the documentation

[canergen]

Put in other fixes, that we discussed when we met. Especially the threshold was computed wrong. Subset_idx is something that I need for resolVI and it needs to be in this function. The argument is also different now with two and three.

[ori-kron-wis]

we need to rerun all tutorials that use DE once merged.

[ori-kron-wis]

this branch was already merged in this PR: #3144