Tophat with --initial-read-mismatches command Error:

[chenyao]

Dear all,

I use tophat with --initial-read-mismatches 10

However I have the error:

gzip: stdout: Broken pipe
error:could not open pipe gzip:left_kept.fa.z

----

I thought it was the gzip error, so I use -z0 to have unzip file, but I have another error:

self.file=open(filename)
IOError: [Errno 2] No such file or directory: '/left_kept_reads_missing.fq'
------------

I use solid pair-end reads. Is it the reason ?

[kmcarr]

chenyao,

You asked in an earlier thread about using the "--bowtie-n" option for TopHat, did you use that option when you got this error? The maximum number of mismatches allowed for -n is 3 so your --initial-read-mismatches 10 would produce an error when the combination of "-n 10" is passed by TopHat to Bowtie. It is possible that TopHat is not trapping this error and then when it tries to open one of the files which should have been produced by Bowtie (but wasn't because Bowtie failed) you get that error.

[chenyao]

chenyao,

You asked in an earlier thread about using the "--bowtie-n" option for TopHat, did you use that option when you got this error? The maximum number of mismatches allowed for -n is 3 so your --initial-read-mismatches 10 would produce an error when the combination of "-n 10" is passed by TopHat to Bowtie. It is possible that TopHat is not trapping this error and then when it tries to open one of the files which should have been produced by Bowtie (but wasn't because Bowtie failed) you get that error.

No, I did not use bowtie-n when I got this error. Also, I got the same error even I set the --initial-read-mismatches to 4.

[kmcarr]

No, I did not use bowtie-n when I got this error. Also, I got the same error even I set the --initial-read-mismatches to 4.

Well again, even for the -v option (default without specifying -n) the maximum allowable read mismatches is 3.

[chenyao]

Well again, even for the -v option (default without specifying -n) the maximum allowable read mismatches is 3.

Are you sure? If it is , then there is no difference between "--initial-read-mismatches" and "--bowtie-n". Why they set two options? And I can't allow the mismatch more than 3? This is unfair for the long read.

[kmcarr]

Are you sure? If it is , then there is no difference between "--initial-read-mismatches" and "--bowtie-n". Why they set two options?

No, they are not the same thing. These two TopHat parameters control the parameters which are passed to Bowtie; "--bowtie-n" runs bowtie with the -n option which specifies counting mismatches only in the initial seed match. The alternative mode is -v which counts mismatches over the entire read length. By default TopHat calls Bowtie with the -v option.

You do NOT pass a number with the --bowtie-n option. By default TopHat runs Bowtie allowing 2 mismatches, in either the full read (-v) or seed (-n) mode. If you wish to changes this for either mode you set it with the "--initial-read-mismatches" option but as stated above this parameter can only accept numbers from 0-3.

And I can't allow the mismatch more than 3? This is unfair for the long read.

Life isn't fair .

If you believe that your reads may contain a larger number of mismatches this when you should employ the --bowtie-n option. Understand that when bowtie is running in -n mode it is only concerned with the number of mismatches in the initial 5' seed.

You should really carefully study the manuals for TopHat and Bowtie to fully understand the run parameters, their limitations and implications for your experiment.

[chenyao]

No, they are not the same thing. These two TopHat parameters control the parameters which are passed to Bowtie; "--bowtie-n" runs bowtie with the -n option which specifies counting mismatches only in the initial seed match. The alternative mode is -v which counts mismatches over the entire read length. By default TopHat calls Bowtie with the -v option.

You do NOT pass a number with the --bowtie-n option. By default TopHat runs Bowtie allowing 2 mismatches, in either the full read (-v) or seed (-n) mode. If you wish to changes this for either mode you set it with the "--initial-read-mismatches" option but as stated above this parameter can only accept numbers from 0-3.



Life isn't fair .

If you believe that your reads may contain a larger number of mismatches this when you should employ the --bowtie-n option. Understand that when bowtie is running in -n mode it is only concerned with the number of mismatches in the initial 5' seed.

You should really carefully study the manuals for TopHat and Bowtie to fully understand the run parameters, their limitations and implications for your experiment.

Thank you so much. So --bowtie-n do not need the number, that's my mistake.