lots of transcript start at 0 and end at 1 in cufflinks output gtf file

Question

[wenhui]

Hi, everyone.

I used Tophat2 + cufflinks +cuffmerge to process my RNA-seq data. The annotation file I used is UCSC Hg38. I saw some strange outcomes form the merged gtf file. There are lots of transcript starting at 0 and end at 1. It really confuses me. Should I keep these data or just trim them?

Any advice is appreciated.

Thank you.

Answer 1

[Charles Plessy]

Long time ago, I tried bowtie/tophat/cufflinks with GENCODE as reference transcript annotation, and also could see a lot of very short transcripts that were created from annotations of codons of interest (start/stop/selenocystein) or nucleotides of interest (polyadenylation sites, ...) in GENCODE. Perhaps there is a similar explanation in your case?

On my side, I stopped using cufflinks. While I have not tried them, perhaps the next generation of tools, namely HiSAT, StringTie or Ballgown may give you better results?

[wenhui]

Thank you for your reply. I have tried hisat2 +stringtie but the software I used for downstream analysis only support gtf file output from cufflinks. It may have something to do with some prefix in cufflinks' gtf file.

Usually, I prefer hisat2 + stringtie in my work and I found it is much faster than Tophat2 + Cufflinks.

Thank you anyway.