Single sample differential expression analysis

Question

[gamma.jian]

Dear all,

I have the following experimental situation: a series of normal samples transcriptomes, and a series of aberrant transcriptomes. However, I don't want to perform the standard analysis Test Vs Normal. Instead, I would like to be able to call for each aberrant sample which genes are deregulated and which genes are not. I know this is fragile from a statistical point of view, having no replicas for my aberrant samples. Can I use standard tools for DEG analysis or is there a strategy particularly amenable to achieve this result?

The aberrant samples are cancers and for the scope of my investigation, I don't want to assume that they are equal. Even if that means losing some power, I want to look at sample specific deregulation comparing that with the distribution in normal samples.

I hope my question is not trivial since I searched for a while and I didn't find a proper answer. Thank you in advance,

Gian

[Devon Ryan]

Are you open to comparing single aberrant vs. all control samples as a group?

[gamma.jian]

Yes, exactly. Control samples are supposed to be homogeneous and thus I'm justified to treat them as replicas. On the other hand, I don't want to apply the same assumption to cancer samples.

[GenoMax]

There is no harm in analyzing data both ways (it will only need time). You may be pleasantly surprised (or badly disappointed), if you only did the analysis the way you want it.

Answer 1

[Devon Ryan]

You can use standard DE tools as long as you have replicates for one of the groups. The power is not going to be great, you'll need more controls, and it's questionable how biologically meaningful the results really are, but it'll work.

[vinaysbharadhwaj]

Hello I want to do single sample DE analysis as well. I have multiple samples and controls and get the DE values for each sample vs all the controls. Could you please help me out by telling how to do so with either limma or deseq2?