maybe I got the purpose of using cuffmerge wrong, but I thought it creates a file including all transcripts that are present in a certain number of samples (e.g. my whole experiment).
I get an output of 97,000 lines running cuffmerge without an annotation file. Using two different .gtf files (NCBI refSeq or UCSC refgene) I get 2.4 million lines or 970,000. How is that possible? And which is the correct one?