Differential Expression Analyses of Entire Connectivity Map (microarray) dataset

Question

[Bioinformatist Newbie]

Hi,

I am interested in finding the differentially expressed genes among all instances of Connectivity Map dataset. CMAP dataset contains a collection of gene expression profiles (6100 instances) that are obtained from cultured human cells treated with various small molecules (approx. 1300). Two kinds of platforms are used in this study GPL96 and GPL3921. There is a separate xls file in supplementary material of their publication which describes the identifiers for vehicle controls and respective treated samples. I want to do differential expression analysis for all instances of every drug vs. respective control. I can do this manually but it will took me months to finish the task. Can anybody suggest me what approach I should follow?

Sometimes arise the problem that if there are 5 GE profiles which are similar drug treated then 2 of them are of GPL96 platform while 3 are of GPL3921, so how I will overcome this barrier?

Thank you for taking out the time to read my post and helping me out.

Answer 1

[Zhilong Jia]

Since I did not do the steps, probably this is a partial answer. There are some articles analysing the cmap data from the beginning, such as Drug repositioning: a machine-learning approach through data integration. There are some points you can get in this article. As the original CMAP article or other documents stated, one of platform is mapped to another one, while unmapped probes are abandoned. About the time you mentioned, these steps can be automatically done by scripts.

However, there are processed amplitude matrix and ratio matrix besides rank matrix concerning CMAP.; Perhaps, what you need is just these data. Read the help of cmap, you will get some information about them.