GTF file generated by stringtie merge has two source ensemble and stringtie

Question

[musk]

Hello,

Can someone help me understanding GTF files generated by stringtie merge. I have downloaded three bam files (repeat) online for a RNA-seq experiment. These bam files were the created by using Tophat. I created GTF file using stringtie with -G option and GTF reference annotation file I downloaded from ensembl. I merged these GTF files (from three bam files) using stringtie --merge with -G option using reference annotation file. The output GTF file has two sources for transcripts; stringtie and ensembl. The transcripts from source ensembl do not even there in original stringtie GTF file that was generated from bam file using Stringtie. Why do we see two sources after merging?

I want to have merged GTF file that has transcripts (novel and known) expressed only in the samples. how will I do that. Any help will be much appreciated.

[lieven.sterck]

I'm not sure what exactly you mean with 'merging' but perhaps you can have a look at the AGAT toolkit AGAT doc . it has functionality to merge GFF/GTF files.

likely you will need to transform you GTF to GFF first

[i.sudbery]

stringtie merge is part of the stringtie suite, and generates a consensus GTF file by comparing multiple assemblies and merging compatible transcripts.

Answer 1

[i.sudbery]

The transcripts marked with an Ensembl source in your output file will have been taken directly from the reference GTF you provided to -G. When you use the -G option, all transcripts from the reference annotation are included in your merged annotation. There is no option I'm aware of in stringtie to filter transcripts from the reference annotation for expression. I far as I'm aware, the filtering options only apply to the transcripts coming from the assemblies you are merging.