search

broadinstitute / lincs-cell-painting

Processed Cell Painting Data for the LINCS Drug Repurposing Project
BSD 3-Clause "New" or "Revised" License
25 stars 13 forks source link

Updating profile & consensus docs #75

Closed michaelbornholdt closed 3 years ago

michaelbornholdt commented 3 years ago

Related to issue #73

michaelbornholdt commented 3 years ago

@shntnu @gwaygenomics @FloHu let's continue the discussion here

michaelbornholdt commented 3 years ago

cytominer_workflow_new

michaelbornholdt commented 3 years ago

I think I still need to add a part where I explain where the profiles and consensus can be found in the repo. That should go in the top level README

michaelbornholdt commented 3 years ago

@gwaygenomics

DMSO vs. Compound Profiles

Note we generated per-well DMSO consensus signatures and per compound-dose pair consensus signatures for compounds. The per-well DMSO profiles can help to assess plate-associated batch effects.

I have read this part of the consensus/README.md several times and I don't understand what it is trying to tell me. Can we rewrite that?

gwaybio commented 3 years ago

I have read this part of the consensus/README.md several times and I don't understand what it is trying to tell me. Can we rewrite that?

Yes! We should clarify anything that isn't clear.

It's just saying that the consensus signatures for DMSO create profiles per well instead of per treatment. Does that make sense? One might assume that we treat DMSO like any other perturbation when creating consensus profiles - that is, combine ALL replicates into a single signature. We don't do that. Instead, we collapse per well, thus creating ~24 or so (I don't remember the exact number) DMSO consensus profiles.

michaelbornholdt commented 3 years ago

@niranjchandrasekaran Can you review and merge this? I don't have any rights here :)

@floHu any last comments?

FloHu commented 3 years ago

@niranjchandrasekaran Can you review and merge this? I don't have any rights here :)

@FloHu any last comments?

I would still clarify what is the difference in the plate normalization and the normalization by spherization, i.e. which artifacts are corrected in the former and which ones in the latter and where they occur. For people from outside the imaging field, e.g. in some screen types people just normalize on a plate by plate basis and that takes care of the batch effects.

Overall, great job, things are a lot clearer now. Thanks for the effort!

gwaybio commented 3 years ago

Thanks @michaelbornholdt - I will provide my review once you address @niranjchandrasekaran's comments

michaelbornholdt commented 3 years ago

@gwaygenomics Ready for you. I addressed Flos and Niranj's comments.

michaelbornholdt commented 3 years ago

@niranjchandrasekaran @gwaygenomics I tried to rewrite in active language and accepted all the suggestions above.

michaelbornholdt commented 3 years ago

Why do the suggestions of Grag still appear in the other tab?

gwaybio commented 3 years ago

@michaelbornholdt - it also looks like you need to add back your new-and-improved workflow figure: profiles/media/cytominer_workflow.png

michaelbornholdt commented 3 years ago

@michaelbornholdt - it also looks like you need to add back your new-and-improved workflow figure: profiles/media/cytominer_workflow.png

The newest version is there!? Or what am I missing?

shntnu commented 3 years ago

Just dropping in to say that this figure was really helpful to me – I had forgotten the order of operations leading up to spherizing and this cleared it up. Thanks again @michaelbornholdt ! 🥇