gcsi drug response table contains duplicate entries

[ymahlich]

Same problem as #455 but for gcsi.
Difference in dataset sizes:

>>> gcsi.experiments.shape
(134250, 8)
>>> gcsi.experiments.drop_duplicates().shape
(133980, 8)

See #455 for more detail.

[jjacobson95]

At a guess, I'd think this might affect all broad/sanger datasets. As a temporary fix, I will manually update the datasets and re-upload, but we should keep these issues open until fixed in the codebase of course.

I'll dm you when datasets are fixed, likely around ~4-5ish.

[jjacobson95]

Just noting, all datasets are fixed (on the data side). This affected more than expected. We need to be more careful in the future for deduplicating rows.

Files affected:

• beataml_samples.csv
• ctrpv2_experiments.tsv.gz
• gcsi_experiments.tsv.gz
• liver_copy_number.csv.gz
• novartis_transcriptomics.csv.gz
• ccle_drugs.tsv.gz
• ctrpv2_mutations.csv.gz
• gcsi_mutations.csv.gz
• liver_transcriptomics.csv.gz
• pancreatic_mutations.csv.gz
• colorectal_transcriptomics.csv.gz
• fimm_drugs.tsv.gz
• gdscv1_drugs.tsv.gz
• nci60_drugs.tsv.gz
• prism_drugs.tsv.gz
• ctrpv2_drugs.tsv.gz
• gcsi_drugs.tsv.gz
• gdscv2_drugs.tsv.gz
• novartis_mutations.csv.gz

Some of these result from whitespace/empty cols.