How can I pass on arguments to bowtie when I run tophat?

[zhiqiang]

Hi,

I want to set the --seed argument for bowtie, but how can I fulfill this when I run tophat? Is there any solution? thanks!

[westerman]

Yes there is a solution. At least for Tophat2. Read the tophat manual. Look for the section that starts with "... TopHat 2 allows users to pass many of these options to Bowtie 2 by preceding the Bowtie 2 option name with ..."

[mgogol]

If you wanted to do something like that with tophat 1, you'd have to edit the tophat script. I think it was nice that they added the option in tophat2 to pass in bowtie parameters.

[zhiqiang]

Yes there is a solution. At least for Tophat2. Read the tophat manual. Look for the section that starts with "... TopHat 2 allows users to pass many of these options to Bowtie 2 by preceding the Bowtie 2 option name with ..."

thanks! I am still work with bowtie 0.12.7 + tophat 1.3.2. Don't know whether tophat 2 can pass --b2-seed to bowtie 2 as --seed.

[zhiqiang]

Yes there is a solution. At least for Tophat2. Read the tophat manual. Look for the section that starts with "... TopHat 2 allows users to pass many of these options to Bowtie 2 by preceding the Bowtie 2 option name with ..."

If you wanted to do something like that with tophat 1, you'd have to edit the tophat script. I think it was nice that they added the option in tophat2 to pass in bowtie parameters.

the random number makes the results not reproducible. You run twice with the same input, but results in different outcomes. This is not suitable for checking results or validating your results with others. That's why I need to set the random number seed.

I edited the tophat script, to add '--seed' in 2 locations.

1. def bowtie():

       bowtie_cmd = [bowtie_path]
        # set seed here
        bowtie_cmd += ["--seed 29"]

        if reads_format == "fastq":
            bowtie_cmd += ["-q"]
        elif reads_format == "fasta":
            bowtie_cmd += ["-f"]

2. def build_juncs_bwt_index():

    bowtie_build_cmd = [prog_path("bowtie-build")]

    # set seed
    bowtie_build_cmd += ["--seed 31"]

    if color:
        bowtie_build_cmd += ["-C"]

but the tophat run failed like this:


[Wed May 16 09:30:47 2012] Reading known junctions from GTF file
Left reads: min. length=75, count=11607291
Right reads: min. length=75, count=11607291
[Wed May 16 09:34:03 2012] Mapping left_kept_reads against genome with Bowtie

gzip: stdout: Broken pipe
[Wed May 16 09:34:03 2012] Processing bowtie hits
[Wed May 16 09:35:15 2012] Mapping left_kept_reads_seg1 against genome with Bowtie (1/3)

gzip: stdout: Broken pipe
[Wed May 16 09:35:15 2012] Mapping left_kept_reads_seg2 against genome with Bowtie (2/3)

gzip: stdout: Broken pipe
[Wed May 16 09:35:15 2012] Mapping left_kept_reads_seg3 against genome with Bowtie (3/3)

gzip: stdout: Broken pipe
[Wed May 16 09:35:15 2012] Mapping right_kept_reads against genome with Bowtie

......

Don't know why.

In addition, I understand setting the seed in this way will lead to every call to bowtie with the same random number within the single tophat run. So I think manage the random number generator in tophat script is better, i.e., in a single tophat run, the multiple callings to bowtie will use a different random number every time.