search

openproblems-bio / openproblems

Formalizing and benchmarking open problems in single-cell genomics
MIT License
287 stars 76 forks source link

rmse -> distance correlation #811

Closed scottgigante-immunai closed 1 year ago

scottgigante-immunai commented 1 year ago

Ping @lazappi @mvinyard . Closes #746

github-actions[bot] commented 1 year ago

Current build status

scottgigante-immunai commented 1 year ago

Going to merge this in a hurry to get results in for the new run. @lazappi @mvinyard if you disagree with how it was done lmk and we can fix post-hoc.

lazappi commented 1 year ago

Code looks fine to me. I'm not sure if anyone has checked with Dan if he had an argument for keeping RMSE. I think he added it originally.

scottgigante-immunai commented 1 year ago

@dburkhardt ^

mvinyard commented 1 year ago

@scottgigante-immunai - I will echo @lazappi in that the code looks good to me. Thanks, Scott! I think I may have added RMSE originally and I have no argument for keeping it as opposed to a correlation metric as you've proposed (which we already discussed over zoom). That said, @dburkhardt may have suggested the metric initially (I can't remember), so it's worth checking. Thanks again!