Strategy for designing small gene panel from large scale transcriptomic experiments

Question

[marion.ryan]

Hello, with regard to a knock-out (transient transfection) in a breast cancer cell line, I am looking for some tips in terms of interrogating the online data bases in a systematic way with the view to generating a smaller gene panel (for transcriptomic analysis) while applying some logic with regard to the targets we choose (so that writing up will be easier). Would you focus on pathways, GO terms, Co-expression networks related to the gene generally or do you look for genes that are co-expressed with the gene being knocked out in specific cell lines? Would you look at genes related to breast cancer for example. Just a few ideas floating in my head there, but reaching out to more experienced users incase I am missing a trick.

[ATpoint]

Can you describe the experimental setup and question a bit more? it is unclear to me what you are trying to do and achieve.

[marion.ryan]

Thanks everyone for your input, really I am trying to obtain a small list of genes to test for perturbations on a Breast cancer cell line following knockdown with a gene called FUCA2. 1) The approach I took in the end was to consider targets downstream of the knockdown gene on pathway. 2) Find an RNASEQ data set (GEO) for the cell line being used, find top 50 genes that were co-expressed with the gene being knocked down (assuming they may be influenced or co-regulated in some way). Created a PPI network on these genes and found that our knock down gene interacted only with one of these genes out of the remaining network. Looked up both of this gene in the context of breast cancer on another database and found that this 'connecting' gene was relevant to to the disease stratification, there were also several hub genes present in this list. Pathway analysis highlighted an over-representation of genes relevant to apoptosis, in this group also, so this is a process I can look at. I also searched the Protein atlas for genes elevated in breast cancer and also in normal breast tissue, performed a GO analysis to examine underlying processes that may be relevant (possibly disrupted) and will; also select some genes from this list, (these would not necessarily be specific to the knockdown gene).

This should give us a list of 30 or so genes, suitable for qPCR and hopefully sound logical when its written up.

Thanks again

Answer 1

[Trivas]

I think you're thinking of this backwards in a way; the "transcriptomic analysis" (in this case, differential expression) IS the way you get your "small gene panel" (activated and repressed genes).