Improved Read Group Handling

galaxyproject / tools-devteam

Contains a set of Galaxy Tools mostly written by the Galaxy Team.

36 stars 92 forks source link

Improved Read Group Handling #92

Closed jmchilton closed 9 years ago

jmchilton commented 9 years ago

BWA/BWA-mem should have an option to use collection information to assign read groups.
Bowtie2 should also have an option to use collection information to assign read groups.
Tophat wrapper should be updated also.
Picard AddOrReplaceReadGroups should have an option to use collection information to assign read groups.
There should be a tool to merge bams in such a way that collection information (sample name) is attached as read groups. (MergeSamFiles? MergeBamAlignment?)

jmchilton commented 9 years ago

Looks like this should be updated to include bowtie2 - my new plan is to create a macro file for read group handling that can be shared between bowite2, bwa, bwa-mem, and I guess other mappers to provide a uniform Galaxy interface - even if the actual command line interface is pretty different.

bgruening commented 9 years ago

Like it, we will probably break workflows but this seems to be worth it.

jgoecks commented 9 years ago

Tophat should be included in this update as well.

jmchilton commented 9 years ago

@jgoecks Added it - thanks!

@bgruening It won't break workflows - it will make upgrading them more difficult :).

natefoo commented 9 years ago

While we're (you're) at it, would a macro for the reference genome selector be possible? The format of that field varies somewhat over tools.

bgruening commented 9 years ago

I'm curious how you will manage this @jmchilton. @natefoo back to my suggestion to collect such snippets separately and make them reusable components

hexylena commented 9 years ago

So...a global macros.xml? yay! Finally!

jmchilton commented 9 years ago

@bgruening My thinking is a macro file that is a library - https://github.com/galaxyproject/galaxy/compare/dev...jmchilton:readgroups?expand=1. I both proud and terrified by this - but if we ignore what is in the macro and just look at how tool authors will consume it - I don't think it is too bad.

As a tool consumer you will need to

Copy of the macro file
Import it
Place either the conditional input set or unconditional list of inputs into your <inputs>
Inject some stuff at the top of your command
Specify the input dataset, multiple input datasets, or input data collection that should be used to automate the specification of this stuff if the user wants to.
Inject another block into command that builds a bunch of variables $rg_id, $rg_lb, etc...
Then build the command - there is even a little utility $format_read_group to help build different command line patterns - should cover a lot of scenarios and prevent the need for a big chain of #if .. #end ifs.

jmchilton commented 9 years ago

@erasche Still skeptical of the global macros - what is generally useful except the stdio tweak - and that should just be a flag <tool fail_reasonably="true"> or something less snarky.

hexylena commented 9 years ago

@jmchilton oh, okay, cool. (I have no horse in this race)

bgruening commented 9 years ago

@jmchilton can we close this one?

jmchilton commented 9 years ago

Sure - tophat still should be updated though.