You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Sometimes, technical issues on either the wet or dry side lead to wells with very few cells. Since all wells are weighted equally when applying normalization and feature reduction (at least, all wells that pass the criteria, whether that's whole plate or something stricter), this has the potential to decrease the quality of normalization since those wells are highly likely to be, well, weird outliers.
You could in theory imagine this being used with any of the pycytominer functions, and to behave in at least two major ways (drop the row entirely vs just ignore it under certain circumstances) - I think my proposed implementation though would be to add it as a property of the SingleCell class (because if you have a cutoff, you presumably want to use the same cutoff all the time) and then use that to ignore failing wells during the scaler calculation in normalize, the feature removal determination steps in feature_select, and the consensus building step in consensus - aka, the well would remain present (but minimally perturbative) up through stage 4 profiles and then only be fully dropped at stage 5. But willing to consider alternative formulations.
Feature example
Alternative Solutions
No response
Additional information
No response
The text was updated successfully, but these errors were encountered:
Feature type
Add new functionality
Change existing functionality
General description of the proposed functionality
Sometimes, technical issues on either the wet or dry side lead to wells with very few cells. Since all wells are weighted equally when applying normalization and feature reduction (at least, all wells that pass the criteria, whether that's whole plate or something stricter), this has the potential to decrease the quality of normalization since those wells are highly likely to be, well, weird outliers.
You could in theory imagine this being used with any of the pycytominer functions, and to behave in at least two major ways (drop the row entirely vs just ignore it under certain circumstances) - I think my proposed implementation though would be to add it as a property of the SingleCell class (because if you have a cutoff, you presumably want to use the same cutoff all the time) and then use that to ignore failing wells during the scaler calculation in
normalize
, the feature removal determination steps infeature_select
, and the consensus building step inconsensus
- aka, the well would remain present (but minimally perturbative) up through stage 4 profiles and then only be fully dropped at stage 5. But willing to consider alternative formulations.Feature example
Alternative Solutions
No response
Additional information
No response
The text was updated successfully, but these errors were encountered: