p-value "NA" in DESeq eventhough counts across the replicates and conditions is not zero

Question

[ashwini]

Hi,

I am getting p-value as "NA" eventhough the log2FC is greater than 6 for the following three transcripts. Can somebody explain me if you have across this anytime?

"id","baseMean","baseMeanA","baseMeanB","foldChange","log2FoldChange","pval","padj"
"ENST00000464588",21567.0703963241,367.539692089608,42766.6011005586,116.359136226658,6.86244068175062,NA,NA
"ENST00000526359",13369.6103749259,109.604047639007,26629.6167022127,242.961982480066,7.92458677529959,NA,NA
"ENST00000532801",12831.1517683669,393.734085139833,25268.569451594,64.1767385788302,6.00397856952259,NA,NA

[Devon Ryan]

This would make more sense if you were using DESeq2, since then these could be genes with apparent outliers. I don't recall there being any mechanism for this in DESeq.

Regardless, you should really be using DESeq2 anyway (it has a good number of improvements).

[ashwini]

Im tried DESeq2 for the same data and this time it is the different set transcripts which are showing "NA" as their p-value

"baseMean","log2FoldChange","lfcSE","stat","pvalue","padj"

"ENST00000315067",789.357858059779,-1.01780504570088,1.33212552574889,-0.764045899599961,NA,NA
"ENST00000342571",111.396126198154,-1.49385254926637,0.668927645604199,-2.23320497976589,NA,NA
"ENST00000373258",228.376850950387,-3.62310290631266,1.50114613047853,-2.41355776946092,NA,NA

The 3 transcripts, where p-value was "NA" for DESeq are showing p-value as 0 with DESeq2

[Devon Ryan]

I'd go with the DESeq2 results. There are known reasons, described in the user manual, why it will sometimes yield NA p-values.

[Martombo]

are you using DESeq or DESeq2?

[ashwini]

I am using DESeq.