Differential Expression Of Protein And Mirna

Question

[dcolinmorgan]

So I've got a new project building off an existing differential gene expression project. My PI wants me to correlate differential gene expression with that of protein and miRNA. I am wholly unfamiliar with techniques for analysis of differential protein and miR expression. I found a paper siting GenePattern from the Broad, and I have used BRB ArrayTools before, but the settings dont seem to be conducive to protein and/or miR datasets.

I am trying to use TCGA data. Any help is appreciated

Answer 1

[Irsan]

Usually, when PI's ask a bio-informatician to do something it turns out they ask the wrong/malformed questions. You should be critical about why correlation (thats how your PI calls it but I think he means association, correlation is just a measure of linearity) are interesting. But assuming they are, I would advice to obtain a matrix with abundance estimates for RNA, miRNA and proteins. To be clear, a matrix is different from differential expression results, it is a table with genes in rows, sample in columns and expression estimates (e.g. FPKM) as values. perform linear regression of miRNA on expression (use log transformed data!) for each gene, test the slope for significance and adjust resulting p-values for multiple testing. You can do the same for any other combination of data types (e.g. expression on protein). In addition to linear regression you can also calculate pearson correlations. These statistical analyses are implemented in the base functionality of R. Additionally, you can use the SIM (statistical integration of microarray data) bioconductor package which uses a similar approach and has some nice additions.

Answer 2

[pld]

These people did a very similar study. http://www.sciencemag.org/content/342/6162/1100.long