Filtering low count when using Tximport from Salmon to DESeq2

Question

[bharata1803]

Hello,

So I've used salmon to extract readcount and use tximport to be used in DESeq2. After I perform DE analysis, there are some genes that has NA result. I suspects that this is caused by low count or maybe even 0 readcount result. Should I filter the low count gene before performing DE analysis using DESeq2 or I can just filter the NA result?

Before, I manually extract salmon result data and make it a matrix and then filter the low readcount. Then I import to DESeq2 using DESeq matrix loader function. With this tximport, I am a bit confused when should I filter the low count.

Answer 1

[h.mon]

When using DESeq2, one doesn't need to filter low counts genes, as DESeq2 performs independent filtering. See the Question: deseq2 filter the low counts BioConductors Forum post for more details.

[Kristoffer Vitting-Seerup]

This is also the reason why there are NAs - because they were removed in the IWH.