Custom reference genomes, bowtie fails to recognize

[Jondeen]

Hi,

I'm trying to install the Schizosaccharomyces Pombe reference genome for use on Galaxy. I've found various instructions on how to do so (it seems routines on using .loc files has changed in recent iterations of Galaxy?), and I see that for bowtie tool-data/bowtie-indices.loc is used for locating reference genome bowtie files. Yet after installing the genome in an accessible location and updating .loc-files (and restarting), the option of applying the S. Pombe genome as reference is still not available.

Specifically the tool I'm interrested in using is Map with Bowtie for Illumina.

What are the most up-to-date instructions on installing reference genomes?

[bgruening]

Hi @Jondeen the recommended way to get your genomes in a reproducible way is to use Data Managers. All you need is to install your favourite Data Manager and run this tool from the Admin panel.

[Jondeen]

Hi,

Thanks. I think I'm on the right track now! :+1: New to Galaxy, and wasn't sure which method was the most up to date.

Thanks for hasty reply!

[Jondeen]

Ooops, sorry to re-open -- maybe this is actually a problem with the specific tool I mentioned, because the off-the-shelf Bowtie2 recognizes the installed reference genome, whereas Map with Bowtie for Illumina does not...

Any pointers as to what I should do?

[Jondeen]

Update:

A s/bowtie-indexes/bowtie2-indexes/g in https://github.com/galaxyproject/tools-devteam/blob/master/tools/bowtie_wrappers/bowtie_wrapper.xml made it show up for me.

[Jondeen]

...but of course that directs bowtie to the bowtie2 indices which does no good. I was missing a data-manager which could generate the bowtie(1) files (*.ebwt) - as far as I can tell there is no such one yet, but calling the generator bowtie-build manually gave me the files and I could install them manually.

[mvdbeek]

There is a data manager for bowtie1 indexes in the testtoolshed: https://testtoolshed.g2.bx.psu.edu/view/jjohnson/data_manager_bowtie_index_builder/119b03a6eaaa this data manage was written by @jj-umn, I believe. Is there any reason it should not be in the main toolshed?
It works very well for us.

[bgruening]

ping the King of data managers @blankenberg! :)

[blankenberg]

It looks OK, but it won't handle the case where you might have multiple different versions of indexes available for a dbkey (e.g. hg19cannonical vs hg19female).