I'm using Cufflinks 2.0.0 without a reference GTF to build models from GSNAP v20120427 alignments, also made without a reference GTF (i.e. using -N 1). I noticed that some of the Cufflinks models extend beyond the end of the reference contig sequence. They also seem to extend beyond the end of the GSNAP alignments as shown in a pileup file.
For example, cufflinks reported this transcript:
But scaffold01237 is only 4696 bases long, and according to the pileup, GSNAP's alignment ends at base 4696.
Is this a known problem with this combination of Cufflinks 2.0.0 and GSNAP 20120427? I can't see how Cufflinks would know to extend the transcript if there isn't data in GSNAP's SAM/BAM output to begin with, and it makes me wonder if Cufflinks is parsing the BAM file incorrectly.
For example, cufflinks reported this transcript:
Code:
scaffold02137 Cufflinks exon 4297 4727 1000 - . gene_id "CUFF.17382"; transcript_id "CUFF.17382.1"; exon_number "2"; FPKM "95.5022086611"; frac "1.000000"; conf_lo "86.719816"; conf_hi "104.284601"; cov "139.246274";
Is this a known problem with this combination of Cufflinks 2.0.0 and GSNAP 20120427? I can't see how Cufflinks would know to extend the transcript if there isn't data in GSNAP's SAM/BAM output to begin with, and it makes me wonder if Cufflinks is parsing the BAM file incorrectly.