Gemma icon indicating copy to clipboard operation
Gemma copied to clipboard

Removal of discrete geeq fields and their replacements

Open oganm opened this issue 1 year ago • 3 comments

@ppavlidis wanted to have the discrete geeq fields removed from the dataset endpoints. However the wrapper packages uses some components, so ideally they would be replaced. These fields are

  • geeq.batchCorrected: boolean field. whether or not batch correction was performed on the dataset.
  • geeq.batchConfound: 0 if no batch info, -1 if batch confound is detected, 1 if batch info is available and no batch confound detected
  • geeq.batchEffect: -1 if p value <0.0001 1 if p value >0.1, 0 for everything else

I believe this information is sufficiently present in batchEffect enum and the newly populated batchEffectStatistics fields. though it's not particularly nice that batchEffectStatistics is free text.

  • geeq.rawData, -1 if no raw data is available, 1 if raw data was available. This is only available by paying another visit to the quantitationTypes endpoing and looking at isRecomputedFromRawData

oganm avatar Mar 29 '24 01:03 oganm

I can relocate some of these fields at the dataset level, where they belong.

The remaining fields can be moved to the admin GEEQ VO.

arteymix avatar Apr 01 '24 18:04 arteymix

We could also simply expose the P-value of the batch effect although right now I think it's mangled in a string.

arteymix avatar Apr 09 '24 16:04 arteymix

that'd actually be nice too

oganm avatar Apr 10 '24 05:04 oganm