I'm trying to run Tophat 2.0.13 (bowtie 1.1.1) on colorspace csfasta and qual files using a GTF produced by Cuffmerge (Cufflinks 2.2.1) as a -G option. Cuffmerge was run on 8 Cufflinked accepted_hits.bam files from Illumina samples.
Tophat crashes and produces an error file m2g_left_kept_reads.err that reads:
"Error at parsing .tlst line (invalid strand):
10772 TCONS_00010773 chr. 118725-119094"
Looking at the GTF file, the problem seems to be lines where the strand is recorded as "." . My first temptation was to delete these lines outright, except that there's 150,000 of them and a big part of what we're looking for is examples of novel splicing- which I'm worried some of these may represent.
I did some searching and I found a few other people complaining about this exact issue, but no solution besides deleting them.
Does anyone know if there's a way to get Tophat to cleanly handle these lines, or a way to change how Cufflinks produces them?
Tophat crashes and produces an error file m2g_left_kept_reads.err that reads:
"Error at parsing .tlst line (invalid strand):
10772 TCONS_00010773 chr. 118725-119094"
Looking at the GTF file, the problem seems to be lines where the strand is recorded as "." . My first temptation was to delete these lines outright, except that there's 150,000 of them and a big part of what we're looking for is examples of novel splicing- which I'm worried some of these may represent.
I did some searching and I found a few other people complaining about this exact issue, but no solution besides deleting them.
Does anyone know if there's a way to get Tophat to cleanly handle these lines, or a way to change how Cufflinks produces them?