Downstream software like the Picard suite and samtools have issues with the values TopHat fills into the fields RNAME, RNEXT, POS and PNEXT for unmapped reads that have a mapped mate. Some errors these tools produce can be seen in this SeqAnswers thread: http://seqanswers.com/forums/showthread.php?t=28155
I've been able to overcome these issues by assigning new values to the fields of the unmapped reads in question:
RNAME: RNAME of the paired read
RNEXT: RNAME of the paired read
POS: POS of the paired read
PNEXT: 0
Additionally, I had to set MAPQ to 0 for all unmapped reads.
Downstream software like the Picard suite and samtools have issues with the values TopHat fills into the fields RNAME, RNEXT, POS and PNEXT for unmapped reads that have a mapped mate. Some errors these tools produce can be seen in this SeqAnswers thread: http://seqanswers.com/forums/showthread.php?t=28155
I've been able to overcome these issues by assigning new values to the fields of the unmapped reads in question:
RNAME: RNAME of the paired read RNEXT: RNAME of the paired read POS: POS of the paired read PNEXT: 0
Additionally, I had to set MAPQ to 0 for all unmapped reads.
These workarounds are implemented in https://github.com/cbrueffer/tophat-recondition , which may speed up verification and testing.