Poissonmultivi

[lauradmartens]

Hi @martinkim0,

I had some time and added the Poisson loss also to the MultiVI model (POISSONMULTIVI) to the external models.

Addresses #2257 for Poisson likelihood.

[lauradmartens]

I added the option to have separate size factors for RNA and ATAC, not sure if this is the best way.

[codecov]

Codecov Report

Attention: Patch coverage is 78.93082% with 134 lines in your changes missing coverage. Please review.

Project coverage is 88.42%. Comparing base (edd1fa1) to head (5e0d1b8).
Report is 117 commits behind head on main.

Files with missing lines	Patch %	Lines
src/scvi/external/poissonmultivi/_model.py	73.58%	70 Missing ⚠️
src/scvi/external/poissonmultivi/_module.py	82.51%	64 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2735      +/-   ##
==========================================
- Coverage   89.40%   88.42%   -0.98%     
==========================================
  Files         161      164       +3     
  Lines       13313    13949     +636     
==========================================
+ Hits        11902    12335     +433     
- Misses       1411     1614     +203     

Files with missing lines	Coverage Δ
src/scvi/_constants.py	100.00% <100.00%> (ø)
src/scvi/external/__init__.py	100.00% <100.00%> (ø)
src/scvi/external/poissonmultivi/__init__.py	100.00% <100.00%> (ø)
src/scvi/external/poissonmultivi/_module.py	82.51% <82.51%> (ø)
src/scvi/external/poissonmultivi/_model.py	73.58% <73.58%> (ø)

... and 2 files with indirect coverage changes

[canergen]

Quick check looks good. Can you explain, why we don't inherit from MULTIVI and MULTIVAE?