featureCounts : limits on multioverlap (option -O)

[marc01]

Hello everybody,

I'm running featureCounts (version 1.6.0) with an annotation in SAF format, this way :

/home/marc/Downloads/subread-1.6.0-Linux-x86_64/bin/featureCounts -F SAF -L -T 3 -O -a saf.test -o /home/marc/Desktop/test_result exon.bam promoter.bam intron.bam five_prime_utr.bam three_prime_utr.bam intergenic.bam

saf.test contains for the 1st run, 50 features, all intentionally on the same region, with the same start & end (overlapping features). The reads were assigned as expected (all the 50 features have the same number of reads).

I have done this test in increasing the number of features, and I saw that for a number of 5000 overlapping features for example, no reads were assigned at all, even though on the screen, I can see it is said that the reads were assigned.

I'm doing these tests, because I want to replace an other tool by featureCounts because of its high speed, and I want to be sure it is free of unexpected results.

So, my question is the following : is there a limit in the number of handled overlapping features by the program ?

Thanks in advance !

Best,

Marc