Calculating RPKM for DESeq

Question

[Whoknows]

Hi friends

DESeq works via count table and create DE genes by count fold change, but it's better to know about gene RPKM values.

Is there any possible way to use / show RPKM for DESeq's output?

And how can i normalized those RPKM values?

tnx

[Sean Davis]

Why is it better to know about RPKM values?

[Whoknows]

I think fold-change where generated based on the count table could be a good measurement, but sometimes read count is not practical (for example those genes with 0 count in one condition) and you have to use another measurement like RPKM and FPKM , Using RPKM make study more robust and confident based on the A.Mortazavi 2008 paper that he said you can detect with 95% confidence if a transcript has RPKM >11.By the way, using RPKM or FPKM let you to decide for choosing only those genes with an exact threshold based on your coverage , and other sequencing criterias.

Answer 1

[Devon Ryan]

Do not mix DESeq(2) and RPKMs. RPKMs are inferior to count data for a variety of reasons. Rethink why you want these values to begin with.

If you want to convert count data to RPKMs, take the normalized counts produced by DESeq2 (or edgeR, or any similar tool), divide them by the gene length in KB (typically one would just sum the exons) and divide by 1 million.

[kanika.151]

Hi,

I want to use rpkm() function in edgeR to do this but i am facing problems with gene length do you know what am I might be doing wrong?

[Chris]

Would you please give code to get gene length and normalized counts produced by DESeq2? I am still new to DESeq2. I appreciate your help.