Re-processing CTRPv2 raw sensitivity data

[MomenehForoutan]

Hi team,

I have been looking at the CTRPv2 sensitivity data here to re-process it myself; however, none of the measurements are actually raw intensity values. For example, their so called "raw data" has been already background corrected and log transformed, and they provide log2 DMSO values per plate (apparently with no background correction as it does not say that in the column description)... so none of these are raw values to start re-processing the data.

Then I see in this page, you say "The CTRPv2 data were generated by the Broad Institute CTD^2 Center and originally released via the Cancer Therapeutics Response Portal (CTRP). The Haibe-Kains Lab has reprocessed and re-annotated the data to maximize overlap with other pharmacogenomic datasets."

So I examined the raw slot of the sensitivity profile in the CTRPv2 PSet, and I do not seem to find anything related to controls, such as DMSO or negative/positive control or blank/media wells. I also went through your Github scripts but cannot find the code you have run to re-process the raw sensitivity data sets.

So based on these, I have a few questions for you:

• Raw intensity data for treatments: Is the raw data you provide as part of the PSet is indeed the raw intensity values? and if so, where did you access that data? Or are you using any of the files listed at the end of this message as they are (e.g. v20.data.per_cpd_well.txt)?
• Raw intensity data for controls: Did you perform background correction and calculated percent to DMSO values? If so, where did you access the raw values for the control wells? Do you provide them as part of the CTRPv2 PSet? I do not seem to find them. Again, are you using the DMSO data as they are from the file listed below (v20.meta.per_assay_plate.txt)? If so, aren't you concerned that the DMSO values are not background corrected while the other raw data (from v20.data.per_cpd_well.txt) are?
• Documentation and code for re-processing: Is your re-processing of the raw sensitivity profiles anywhere available that I can read? That would be super helpful if they are available anywhere.

Many thanks in advance for your time, and looking forward to hearing from you soon.
Best wishes,
Sepideh

---

Here are some details I could extract from the data they provide on the website:

File v20.data.per_cpd_well.txt: raw and transformed viability values for each cancer cell line treated with compound for each concentration point tested for each replicate tested

• Column raw_value_log2: logarithm (base 2) of baseline-corrected chemiluminescence value: log2(max[raw_value - baseline_value, baseline_value])
• Column bsub_value_log2 - log2-transformed compound-well value background-subracted using dmso_plate_avg_log2

File v20.data.per_cpd_avg.txt: transformed and averaged viability values and error-propagation statistics for each cancer cell line treated with compound for each concentration point tested

• Column cpd_expt_avg_log2 - weighted average of replicate bsub_value_log2; shared by all replicates of a master_cpd_id + cpd_conc_umol + experiment_id

File v20.meta.per_assay_plate.txt

• Column dmso_plate_avg_log2 - per-plate mean of log2-transformed DMSO-well values

File v20.meta.per_experiment.txt

• Column baseline_signal - chemiluminescence signal of wells with media but no cells