How many mismatches does Tophat allow?

[wudinger]

Hi, guys:

Does anyone know the maximum mismatches that tophat allows for each mapped RNA-seq read? I recently run tophat with 45bp RNA-seq reads and then convert the result into sam format with MD tag. Some results are like this:

HWI-EAS412:2:1:88:10751:10335:0 147 Muller-A 1576 255 46M = 1541 0 GGTTGGCGGCGCCGGGGCGGCAGCCTGCCCGGACCTTGTGCTTTTC ##############AB05/7,-?A->BDD=D?3B*B+;457ABA:A NM:i:7 NH:i:1 MD:Z:1C0G1T2T5T4A19G7

'NM:i:7' means there are 7 mismatches, which is way more too much for a RNA-seq read? Does anyone know about that? Thanks!
my command is:

tophat -p 8 --coverage-search --microexon-search --solexa1.3-quals --keep-tmp --mate-s
td-dev 50 --mate-inner-dist -30 ...

[Gonza]

I think the default is 2 mismatches. I do not know if you would want to change that to 0 ???

Options:
-v/--version
-o/--output-dir <string> [ default: ./tophat_out ]
--bowtie1 [ default: bowtie2 ]
-N/--read-mismatches <int> [ default: 2 ]
--read-gap-length <int> [ default: 2 ]
--read-edit-dist <int> [ default: 2 ]

[blancha]

I think you have a bigger problem.
The quality of some of your bases seems extremely low.
You could trim, but 45 bases is already quite short.

I would trim the lowest quality bases, and repeat the alignment. But, the reads would then be so short, that the alignments would not be unique.
I think you just have bad quality data.
I wouldn't fault TopHat.