Complete our short form to continue

Genestack will process your personal data in accordance to its privacy policy which can be found here. This includes sending you updates by email about our products and content we think it would be of interest to you. You can unsubscribe at any time by clicking the link in the footer of any email we send. By clicking submit you agree that we process your information in accordance with these terms.
Archive

Using Genestack for pharmacogenomics research

23.10.14
pharmacogenomics app model

My name is Roma Kurilov, I study in Heidelberg at the Helmholtz International Graduate School for Cancer Research at the German Cancer Research Center (DKFZ). I have always been interested in molecular biology: since doing my bachelor's degree I have worked on various wet lab and computational projects in St. Petersburg and Heidelberg. In this article I will describe how my PhD research project takes advantage of Genestack's Genomics Operating System to try to make some pharmacogenomic predictions.

Recently I started my PhD project in Applied Bioinformatics group of DKFZ (German Cancer Research Center) led by Professor Benedikt Brors. The idea of our research lies in the field of pharmacogenomics: we are working on the prediction of drug-response in cancer patients, building a model based on the patient's personal genomic traits. This project is based on a large microarray dataset of cancer cell lines. Genestack supports microarray data, and as the first step I developed two new Genestack applications for analysing expression arrays and SNP arrays. The first one calculates normalized expression values for the given experiment and provides quality metrics informations using the arrayQualityMetrics Bioconductor package.

The second application performs Copy Number Variation analysis for SNP arrays, calculating Copy Number estimates for the probes in the given experiment. This app is based on the crlmm Bioconductor package.

Currently I am working on my third application, which will combine the outputs of the previous two applications for expression and CNV analysis together with a database of cell line-specific drug-response values (IC50). The application will create regression models for IC50 values prediction:

romakur-formula

As a starting point we plan to use for our model training the data from two big public pharmacogenomics resources: Cancer Cell Line Encyclopedia and Genomics of Drug Sensitivity in Cancer. Hopefully, in the near future I'll report about first modelling results on the pages of this blog.

Let's keep in touch,

Roma Kurilov, MSc

23.10.14

Sign up for our newsletter