why cuffmerge need a fasta sequence as running parameter

Question

[ibphuangchen]

Hi there,

I've got a question when I go through the tuxedo RNA-seq data processing, -- why cuffmerge need a fasta sequence as running parameter (i.e. -s genome.fa)?

I would imagine that what cuffmerge does is to further assemble transcripts that have been "preliminarily assembled' by cufflinks, by integrating all those transcripts.gtf files generated from cufflinks. GTF files have got all the genome coordinates information, and the resulted GTF file from cuffmerge has only output genome coordinates. So, why we need to put a fasta genome there when we run cuffmerge?

Thanks in advance.