Hi,

I'm trying to use the TissueEnrich package to determine which genes are differentially responsive to a drug treatment and I'm not quite sure how to get the data into a format that will work for TissueEnrich. It appears that TissueEnrich requires a table with TPM or FPKM for each tissue. However what I have is 4 biological replicates for each sample, so do you know how I would indicate that all of these samples are the same tissue? The other question (which is maybe more important) is can TissueEnrich work with differential expression results? For example, in our experiment we have treated each sample with either drug or vehicle and we're interested in knowing whether tissues behave differently in response to the drug. So our metric rather than being TPM or FPKM, which are absolute expression levels, would be a log fold ratio of expression in drug treatment / vehicle treatment, which will include some negative values. Will that work for TissueEnrich, or if not then do you have other suggestions for applying our data to TissueEnrich?

I read the vignette here and the manual but didn't see any reference to dealing with replicates or differential expression. In the vignette, it does say that for the example dataset, "In order to make the tissue-specific gene calculations more robust, we only used tissues that had ≥2 biological replicates." However it doesn't say how these replicates were handled. Maybe the data for the sample replicates was merged?

Thanks,

Andrew

Hi Andrew,

As from your description, it seems like you are trying to find the differentially expressed genes based on the drug treatment. I recommend that you should use tools like DESeq2 or edgeR for that. Also, we took care of replicates by taking the average of the gene expression across replicates. TissueEnrich in its implementation only works with positive values. One way you can use TissueEnrich is to take the absolute fold-change values.

Regards, Ashish Jain