On 2015 Sep 30, John Quackenbush commented:

My colleagues and I have conducted an expanded analysis of the CCLE and CGP/GDSC data, including additional data that these groups have released since 2012. Our goal in doing this analysis was to address some questions that people had raised about how we measured correlation between reported drug response. For example, one would expect highly targeted therapies may have many non-responsive cell lines representing phenotypic noise where one might not expect correlation. Could smarter filtering of the phenotype data uncover missed correlations? Would binary characterization of the cell lines as either responsive or nonresponsive improve the observed correlation? Would other correlations metrics better measure drug response correlations? The results largely support our initial findings of a general inconsistency and suggest the need for standardization of how drug response is measured.

Our updated analysis is available on the bioRXiv. We welcome comments as we move this work to publication.

On 2015 Sep 30, John Quackenbush commented:

My colleagues and I have conducted an expanded analysis of the CCLE and CGP/GDSC data, including additional data that these groups have released since 2012. Our goal in doing this analysis was to address some questions that people had raised about how we measured correlation between reported drug response. For example, one would expect highly targeted therapies may have many non-responsive cell lines representing phenotypic noise where one might not expect correlation. Could smarter filtering of the phenotype data uncover missed correlations? Would binary characterization of the cell lines as either responsive or nonresponsive improve the observed correlation? Would other correlations metrics better measure drug response correlations? The results largely support our initial findings of a general inconsistency and suggest the need for standardization of how drug response is measured.

Our updated analysis is available on the bioRXiv. We welcome comments as we move this work to publication.