[BioC] Tool inqured to analysis array expression and cell line screening data

Steve Lianoglou lianoglou.steve at gene.com
Mon Jan 27 18:41:06 CET 2014


On Sun, Jan 26, 2014 at 9:01 PM, Ran [guest] <guest at bioconductor.org> wrote:
> I would like your suggestion for the bioinformatics approaches how to select biomarker genes in the cancer cell lines. We have IC50 data for a compounds screening to about 100 cell line, resulted in compound sensitive or resistant for the cell lines. We would like to have tool to select gene or marker correlated with the gene expression or copy number changes. How can we use the public data (e.g. CCLE’s gene expression etc.) to select the genes or biomarkers – which correlate with the IC50 screening for the cell line?  Thank you.

The most obvious thing to do is to just mimic what they reported to
have done in their original publication:

Systematic identification of genomic markers of drug sensitivity in cancer cells

They combined different types of "omics" measurements from the cell
lines they have drug response data from and used them as features for
their favorite machine learning algorithm (the elastic net (implented
in the glmnet package on CRAN)) to predict drug response (measured in
a few different ways).

It's easy to say. Easy "enough" to do. Difficult to do correctly, but
I will leave it at that.


Steve Lianoglou
Computational Biologist

More information about the Bioconductor mailing list