Example Data¶

ClimbTheCliff ships with the in-vitro benchmark dataset from the manuscript: six bulk RNA-seq mixtures (BULK1-BULK6) pooling four human leukemia cell lines (HL60, K562, SUDHL4, THP1) in known, varying proportions, together with a matching single-cell RNA-seq reference and measured drug sensitivity. All three objects are lazy-loaded with data() once the package is attached.

Object	Description	Manuscript figure
`sc.es`	scRNA-seq reference, the 4 cell lines profiled individually (2342 cells)	Fig. 2
`bulk.es`	Bulk RNA-seq of the 6 mixtures, with true mixing proportions in `pData(bulk.es)`	Fig. 2
`bulk_auc_data`	Drug sensitivity (AUC) of the 6 mixtures for 7 drugs x 3 timepoints (`sample`, `drug`, `auc` columns, ready for `cliff()`)	Fig. 4A, validated against pure cell-line viability in SFig. 7A

library(Biobase)
data(sc.es); data(bulk.es); data(bulk_auc_data)

table(sc.es$cellType)               # cells per cell line in the reference
pData(bulk.es)                      # true mixing proportions per bulk sample
unique(bulk_auc_data$drug)          # available drug x timepoint combinations

bulk_auc_data$drug combines each of the 7 drugs (5Aza2d, Cytarabine, Doxorubicin, Imatinib, Mitomycin, Nilotinib, Rapamycin) with each of 3 timepoints (48h, 72h, 96h), e.g. "Cytarabine_48h". cliff() takes one auc column at a time, so subset to a single drug level before calling it — see the examples in CLIMB reference and CLIFF reference.