Include new data sets

[PierreBoyeau]

• pbmc_new (not sure it is here right now)
• 1k1k
• IBD: talk to Can
• Sciplex: @justjhong
• Drug discovery: @adamgayoso

[adamgayoso]

https://www.sciencedirect.com/science/article/pii/S0092867422014702

[PierreBoyeau]

From the paper: "We next evaluated whether nuclear hashing could enable chemical screens by labeling cells that had undergone a specific perturbation, followed by single-cell transcriptional profiling as a high-content phenotypic assay. We exposed A549, a human lung adenocarcinoma cell line, to one of four compounds: dexamethasone (a corticosteroid agonist), nutlin-3a (a p53-Mdm2 antagonist), BMS-345541 (an inhibitor of nuclear factor κB–dependent transcription), or vorinostat [suberoylanilide hydroxamic acid (SAHA), an HDAC inhibitor], for 24 hours across seven doses in triplicate for a total of 84 drug–dose–replicate combinations and additional vehicle controls"

@justjhong do you have an idea how we should model samples for this data?

[justjhong]

@PierreBoyeau my approach had been modeling each combination of drug, dose, replicate as its own sample, but this resulted in the number of samples exploding and there being few observations per sample. Other than subsetting the number of samples, I didn't find a better approach