Decrease the low counts

Question

[rufusek500]

Hello, I am a beginner when it comes to R and I'm still learning how to use R. Currently I'm performing the analysis, where I have to perform both the normalization DEseq2 and clustering. I've done the normalization and I obtained the following results:

out of 40722 with nonzero total read count
adjusted p-value < 0.05
LFC > 0 (up)       : 19, 0.047%
LFC < 0 (down)     : 0, 0%
outliers [1]       : 0, 0%
low counts [2]     : 11537, 28%
(mean count < 0)
[1] see 'cooksCutoff' argument of ?results
[2] see 'independentFiltering' argument of ?results 

My aim is to decrease low counts from 11537 to 2000, but I have no clue which the function or command I should use. I was wondering regarding normalize.quantiles, but I'm not sure. Do you have any hints or suggestions for me?

Thanks in advice

[iraun]

Hi @rufusek500, please consider editing your answer using the available formating options so it becomes more readable :).

Answer 1

[swbarnes2]

You've got 40k entries in your DESeq object. That's about all the genes? Well, if 20% of all genes are not expressed in your samples, then those genes should have low counts. This isn't a problem to fix, it's reality. You can't fix it, and you shouldn't try. (Especially with DESeq2, it wants raw counts only!) The only thing you can do is to remove those genes early in the process. There are a couple ways to do this in the vignette, under "Pre-filtering"

[rufusek500]

Yeah, you are right, thanks a lot for your help :)