Dear everyone,
I'm sorry for the following question which is more "understanding" than technical but...
I'm studying the function of one gene, Otx2. I have access to 2 RNA-seq experiments
One sequencing of the Choroid plexus w/ 4th and lateral ventricle (Control vs siRNA Otx2; 2 biological replicates each time, 8 samples total) 50 bp single-end
One sequencing of the cortex (Control vs Heterozygote Otx2 +/-, 2 different time points, 2 biological replicates, 8 samples total) 50 bp single-end
I already performed for each DEG, and alternative splicing analysis.
However, what I really want to find is the pathways regulated by my protein, and I find that simply overlapping the list of DEG's is quite inaccurate because I lose all the information about expression and more importantly, maybe there is relevant information in the >0,05 p-value. Same when I do pathways analysis (STRING) I loose differential strenght (I mean, it is not the same for the 1st and last gene of my list, but string considers the two equal..).
I wanted to perform WGCNA but it seems that my sample number is too low (apparently 15 minimum) Do you think it is relevant to mix my data? Is there other ways to analyze these datasets ? I would love to mix DEG and Differential isoform usage as well !
I hope someone can help me, I know it seems very demanding, but I'm learning bio-informatic all by myself and it's quite challenging sometimes...
Thank you very much!
David
If you don't want to lose "information about expression", don't do an overrepresentation analysis, rather do a GSEA analysis for GO terms, pathways, etc. In GSEA, you input a ranked list of genes (ranking your genes based on strength of differential expression).
Thank you, i'll dig into this, I didn't know about it. Do you think it is possible to mix my two tissues with it in order to find pathways that are only dependant on my protein ?
I mean, yes, you can do GSEA for each tissue, and compare the enrichment results.