Hi, I am working to perform a Ribo-IP analysis, and I have some problem in calculating translation efficiency and my dataset is different from the traditional Ribo-seq. Here is my dataset:

file       assay   group
A_IP_1     IP      A
A_IP_2     IP      A
A_WT_1     WT      A
A_WT_2     WT      A
A_RNA_1    RNA     A
A_RNA_2    RNA     A
B_IP_1     IP      B
B_IP_2     IP      B
B_WT_1     WT      B
B_WT_2     WT      B
B_RNA_1    RNA     B
B_RNA_2    RNA     B

IP means RIbo-IP , WT means empty-IP(beads cannot capture Ribosome-protected RNA but some background RNA binding on beads), RNA means the total RNA(RNA-seq)

What I would like to do: (1) calculate translation efficiency(TE) of A and B (2) calculate the differential TE between A and B

I have known that I could use interaction term to calculate ratio of ratio, but my situation is a little different. I found there are many small non-coding RNA reads in my IP samples, and these reads were also enriched in WT samples, so I think these reads are from contaminated RNA which binds on beads. If I use the traditional method to calculate TE, like IP/input, the gene with highest TE are these non-coding genes.

My main question is: how can I remove these contamination reads when calculating TE and differential TE? Could I calculate TE by (IP-WT)/input or IP/WT/input?

Thank you for your time and help,

Andy

Why don't you simply remove anything from the count matrix that you think should not be there, so removing the rows that you do not want?