Does STAR aligner need the information from annotation file to identify chimeric reads?

[Xiao Lei]

Dear All,

I wonder if the STAR aligner can identify chimeric reads just by using the information from the hg38 genome (sequence only) without any information from annotations. 

Thanks, 

Xiao

[Alexander Dobin]

Hi Xiao,

Yes, STAR chimeric detection does not use annotations, so you can chimeras between unannotated transcripts.

[Xiao Lei]

Hi, Dr. Dobin,

Thank you very much for your reply! To create an index of a genome (e.g. hg38) for STAR alignment without annotations, my understanding is that we just supply the genome for the index process, if we already indexed the genome together with annotations (e.g. hg38 with GTF or GFF3 annotation files), do you mean even in this case STAR detects chimeric reads without considering annotations?

Regards,

Xiao

[Alexander Dobin]

Hi Xiao,

Chimeric junction detection does *not* depend on annotations, independent of whether the genome was generated with or without annotations.

If you provide annotations, STAR will get better alignment to annotated (normal) splice junctions. This will improve chimeric alignments between annotated transcripts.

Cheers

Alex