low number of predicted m6A sites

[baibhav-bioinfo]

Hello,
I have nanopore DRS data for my plant sample. My samples are sequenced using RNA004 chemistry.
But as there is non-availability of RNA004 based model for plant, i used the RNA002 based arabidopsis model for the prediction using the "inference" command.

The number of predicted sites after >0.9 (prob_modified) cutoff is very low (3000 per sample) is very concerning to me.
As i have almost 15 million DRS reads per sample.

is this by chance/true or anything wrong in my analysis?

[yuukiiwa]

Hi @baibhav-bioinfo,

RNA002 and RNA004 chemistries differ quite a bit, so the RNA002 model is unlikely able to detect m6A sites in RNA004 samples (we tested this on human). We do not have an RNA004 Arabidopsis model.

Thanks!

Best wishes,
Yuk Kei