Receiving more reads from exons than whole genes using featurecounts

Question

[till.schwaemmle]

Hi,

In order to receive intronic counts for my rnaseq data, I wanted to substract the reads for exons from the total reads of the respective genes. For this, I am using featureCounts from the RSubread package in R. However, when I substracted the reads, some genes were returning negative counts (meaning there are more reads in the exons than in the whole gene). As a .gtf file I am using the standard mm10 annotation from ensembl with the standard options with GTF.featureType set to "gene" or "exon" respectively, except that I set fracOverlap to 1 (to remove any intronic reads from the exon reads). Does anybody have an idea how this can happen?

Thanks in advance