Considering that the basal expression level is not 0 for the calculation of the differential expression

Question

[Pin.Bioinf]

Hello,

I am doing a RNA-Seq DE analysis with DESeq2 and I was told to not consider the gene basal expression level as 0, so, how would I do that step? Are they referring to the relevel function in DESeq2? or setting the control samples as the targets of the comparison? Or is there more to it that i am not aware of?

Also, would results differ a lot if I do the analysis as indicated in the DESeq2 tutorial and not do the "not considering basal expression level as 0" step ?

Thank you

[Joe]

I think this has to do with calculating fold changes and log changes. With basal levels of 0, you would end up dividing my zero and/or taking logs of 0s, both of which will give you infinite values.

[Kevin Blighe]

and I was told to not consider the gene basal expression level as 0

Who told you?

Are you referring to zero raw counts or 0 normalised counts?

[Pin.Bioinf]

My biologist fellow. Normalized counts

[Kevin Blighe]

You should obtain the logged counts and then covert these to the Z scale. Then work on expressed / not expressed based on the Z scores.

[Pin.Bioinf]

Logged counts? You mean the log2 of my normalized data and then transform to Z scale?

[Kevin Blighe]

Yep - that's it.