Entering edit mode
2 days ago
aLex97
•
0
I had RnaSeq data and I've got read matrix from rsem. Now I want to remove batch effect. I decided to do that with ruv seq (RUVg)
What is the best approach to choose negative control genes for RUVg (based on my data)?
It is generally expected in communities like this that you read the manual first which covers this https://bioconductor.org/packages/release/bioc/vignettes/RUVSeq/inst/doc/RUVSeq.html#empirical-control-genes and then tell what beyond this poses a challenge.
I've tried to intersect the DEGs from edgeR with the compared groups (genes that didn’t meet the criteria to be included as DEGs, such as p < 0.05 and high FDR). However, I didn’t get satisfactory results, probably because I lost some genes in that comparison.
Next, I normalized the matrix and focused on genes with low variance (below 0.2). This approach gave me a list of about 1500 genes, but I need to verify this by checking the PCA plot after removing batch effects.
I’m asking if anyone with more experience could share advice, as I don’t have a strict method to validate results I get.