Subread low percentage of assigned fragments in ENCODE data

[kajot]

I found the problem, I was using wrong GTF annotation, should've used GENCODE mouse M1 since vM3 is built on mm10...

[shi]

Could you show the content of the ".summary" file which was part of featureCounts output?

[kajot]

Hi there guys!


I am using Subread v1.5.0-p1 to get tables of counts for RNAseq data I got from ENCODE (https://www.encodeproject.org/experiments/ENCSR000BYX/).
What I am trying to get are counts for genes (using reads assigned to exons) since I want to ultimately turn the numbers into FPKM or TPM.

However, I am getting around 13% of successfully assigned reads:

|| Total fragments : 72502756 ||
|| Successfully assigned fragments : 9324091 (12.9%) ||
|| Running time : 2.20 minutes ||

This doesn't sound right to me, I am used to getting 50-75% of reads assigned in RNAseq, using HTseq + DESeq2. I don't expect ENCODE data to be of poor quality so I think it's something on my side.

I tried all the combinations of fr-firststrand, fr-secondstrand, rf-firststrand, rf-secondstrand etc. Nothing gives me higher values than those 13%.
Using GENCODE M3 annotation gtf since the data was originally aligned to mm9.
This is a paired-end library from Illumina GAIIx sequencer.

Any tips ?