View Single Post
Old 02-27-2014, 12:38 AM   #7
Simon Anders
Senior Member
 
Location: Heidelberg, Germany

Join Date: Feb 2010
Posts: 994
Default

Does your GTF file contain "exon" lines for all the anti-sense transcripts? If so, this would explain the discrepancies.

htseq-count considers a read as "ambiguous" if it maps to more than one feature (gene). So, if a read maps to a position which is covered by a regular gene on one strand and by an overlapping antisense transcript on the other strand, then this read will be counted as ambiguous if you have set "stranded" to "no", because there is no information to decide whether the read originated from the sense of from the antisense transcript. For "stranded=yes", however, the read will be counted for the feature that is on the same strand as the read.
Simon Anders is offline   Reply With Quote