RNASeq data RPKM

Question

[tanyabioinfo]

Hi

I have RNA Seq (RPKM) data. I am trying to look for fold change in the gene expression for some genes.

Say the RPKM value in condition1 is A and for condition 2 is B

For few genes I have A as zero values. So when I calculate fold change(B/A). For the genes which has A as zero the value would become infinite. Question1: How can the infinity value be taken care of. Is Adding 1 to all the data a good way to handle it, specially when I am interested in fold change. Question2: Is the fold change as B/A correct of doing it.

Best

Tanya

[Ron]

Yes adding 1 to FPKMS is a good way to take care of this. The fold change of B/A is fine but you have to take log of the fold change.

[tanyabioinfo]

Hi Hussain

Thanks for your reply.

I have this data at three time point with the ligand binding at the second and third time point. I want to see the effect of the ligand on gene expression. I am looking at overall pattern of the gene so I need to consider these cases. Do you think adding 1 to the entire dataset a good way to overcome such issues.

Best

[Hussain Ather]

Adding 1 would overcome the issues.

[GenoMax]

tanyabioinfo : Please use ADD COMMENT/ADD REPLY when responding to existing posts to keep threads logically organized.

This comment should have gone under @Hussain's answer.

Answer 1

[Hussain Ather]

You can't compute fold change if the gene wasn't detected. Ignore them. The fold change is B/A.