Entering edit mode

17 months ago

grayapply2009
▴
240

I normalized my raw counts using Voom (limma package) and got a strange Mean-Variance plot https://ibb.co/C2RZXGk . It doesn't look right.

Any suggestions about why and how to deal with it?

That does indeed look very stange! How did you quantify your samples? How many samples do you have? Did you do any filtering before using voom - if so how?

Thanks for your reply. Here is what I did: 1. Quantification: fastq -> count table (Star + featurecounts, mm10, N = 30) 2. Remove lowly expressed genes: genes showing less than or equal to 1 count in 30% samples were removed 3. Voom (TMM) normalization

The following plot is the voom normalization without removing any genes (https://ibb.co/6N8by9B).

Probably this isn't the problem, but the variance is (standard deviation)^2 while you are plotting sqrt(standard deviation).