Hello,
I am working on a non-model fungal organism and I am trying to assemble transcripts with StringTie based on RNA-seq reads mapped to the assembled genome. My ultimate goal is to perform differential expression analysis at the transcript level. However, I noticed some cases where StringTie merged transcripts from neighbor genes (see here). I suspected this issue was due to gene fusion in the genome annotation, which was used within StringTie to perform annotation-guided assembly (parameter -G). However, even without the annotation, StringTie still merges neighbor transcripts.
Fungal genomes are compact, and overlapping UTRs of neighbor genes is not a rare phenomenon. Can someone suggest a way to minimize these assembled chimeric transcripts?
I appreciate any help. Thank you
Seems like erroneously predicted UTRs could be a problem. The reconstructed transcripts look better when I removed the gene UTRs from the annotation (shown here).