Entering edit mode
7.1 years ago
User000
▴
690
Hello,
I have log normalised counts of 50K expressed genes and as conditions I am using 10 different sorts of that fruit. I need to compare expression of sorts; i.e. how much does gene expression variate between sorts? And I have to estimate the variance for each condition. Any suggestions are much appreciated. I am using R and deseq2.
EDITED: So, I have 5 tissues for 10 sorts, I want to compute the variance per gene of tissues and I want to consider my sorts as reps. Does it make sense? May be I can divide my table in 5 and do smth like this? var.per.gene <- apply(log.norm.counts, 1, var)
Do you have replicates? And raw data?
Yes, I have raw counts, no replicates, although I have different tissues( can I treat them as reps in this case?)
Good, because DESeq uses raw counts as input (and not normalized counts). A useful workflow can be found on here.
Not good, because (biological) replicates are important. I'm unfamiliar with fruits, but commonly different tissues are not acceptable as replicates because the variability between tissues is large (and often larger than the biological effect you want to investigate.)
thank you very much, I already used deseq2 to normalize my raw counts and plotted PCA. So now I would like to compute the variance per gene of tissues. I have seen the workflow several times but haven't found something that could answer my question, if I missed something could you please indicate in which section. Many thanks