Understanding "filter counts"

Question

[ovariohisterectomia]

hi everyone, im pretty new in R and in RNA-Seq, im writing my script for DE analysis. I have read about the filter of low counts, and i have a doubt about rowSums Counts. This is part of my script, that a teacher sent me, and im not understanding what does mean the number 2 and the number 12... i would apprecieate a "for dummies" answer.

Im asking mainly because i made my PCA plot and despite are replicates (2 groups, liver and lung, and 6 replicates per group) each sample seems to be very long

I have read that this should not influence the PCA (and that it is supposed to be correct. Worst could be that a variation in this numbers that im asking, changes the PCA)

To sum up, what does that numbers mean and should this numbers modify my PCA?

Thank you everyone!

### Filter low counts ### 

> dds <- estimateSizeFactors(dds) 
> idx <- rowSums( counts(dds, normalized=TRUE) >= 2 ) >= 12  
> dds <- dds[idx,]

[GenoMax]

I tidied up your code a bit. Make sure it still looks correct.

Please use the formatting bar (especially the code option) to present your post better. I've done it for you this time.

Thank you!

Answer 1

[ATpoint]

This code snipped is checking which genes have more than 2 normalized counts in >=12 (so all) samples.

Please see the manual towards filtering counts for DESeq2. PCA should not be influenced by this as plotPCA uses the most variable genes and these you will not find among those with super-low counts across all conditions.

[ovariohisterectomia]

thank you so much, i was not understanding this:

This code snipped is checking which genes have more than 2 normalized counts in >=12 (so all) samples.

Now i get it, thank you!