I first posted this over on SeqAnswers, but figured cross-posting may be ok.
My lab is studying a cell population in Drosophila that is difficult to isolate, but genetically tractable. We created some flies in which we have genetically ablated the cells of interest using the GAL4/UAS system to induce apoptosis. We have collected control and experimental samples in triplicate and performed a paired-end RNA-Seq experiment using the Illumina HiSeq platform.
We are now analyzing the data. My question is whether anyone has seen examples of this kind of approach before.
Since the cells in question make up only a small proportion of the isolated tissue, we expect that most of the down-regulated genes will be fairly specific to this particular cell type. Up-regulated genes will likely be related to the induction of apoptosis itself and the process of cleaning up the mess.
Has anyone tried this sort of approach on cell types whose expression profiles are otherwise difficult to assess.
I have tried searching Google Scholar for as many search terms as I can think of to identify papers doing something like this, but I have so far come up empty. I guess this either means I am a pioneer or an idiot.
Any comments or references would be helpful. Also warnings on things we will need to be aware of during data analysis.
This is so vague as to not be very helpful, but you could try looking at the cancer genomics literature. They have a similar problem where the samples they analyze are mixed (tumor and non-tumor, etc.). And this won't help you analyze your current data, but you could start thinking about single-cell RNA-seq for the future. It's becoming more common and there are kits available from several vendors.
Thanks matted. Sorry for being vague. I just didn't know how much detail would be helpful and how much would be more information than is necessary. I guess people don't read questions that get too long-winded.
I am happy to give any more details that would help us to analyze these data properly.
Sorry, I wasn't clear, I meant my response and suggestions were vague! I apologize if it came off as negative towards your question.
Another method, considering that this is expression data, is to try building co-expression networks. If the genes are working together, e.g. within the same cell type?, maybe they will form a gene cluster. One tool that I can think of will be WGCNA. Maybe you can give it a try
I haven't tried anything like that yet, but I will give it a shot. Thanks Sam.