On 06/06/13 02:58 PM, Björn Grüning wrote:
Am Donnerstag, den 06.06.2013, 14:42 -0400 schrieb Sébastien Boisvert:
On 06/06/13 02:32 PM, James Taylor wrote:
I want to add Ray (a scalable de novo assembler for genomes and metagenomes) to Galaxy.
And I really want you to do this!
So we are at least two !
Joining the party! Make three out of it!
I will also have to write a wrapper for Ray to prepare the command line from the options provided by the Galaxy API.
But where is stored the executable (in my case, where is sdtored Ray) ?
Does Galaxy include the specs to build all the tools available in Galaxy-Central ?
No, we are gradually moving all the tools out of galaxy-central into the Tool Shed. You probably want to look at this page:
http://wiki.galaxyproject.org/ToolShedToolFeatures#Automatic_third-party_too...
Which describes how you can add a recipe to the toolshed that will install the Ray binaries.
So Ray needs a C++ 1998 compiler and a MPI library (let's say g++ and openmpi-bin). Are those already installed in Galaxy deployments ?
You can assume that gcc is installed. Nevertheless, openmpi-bin needs to be installed. I would suggest to create a orphan-tool-dependency for openmpi-bin. For example have a look at: http://testtoolshed.g2.bx.psu.edu/view/bgruening/package_boost_1_53
And than make that orphan-tool-dep a requirement of your repository.
Also, when is tool_dependencies.xml processed (I assume it is procesed when re-loading the content of the shed into Galaxy ?
I don't understand the question,
On some platforms (like DNANexus), when you launch an app, it launches an instance from an image (let's say Debian). Then, the system install the software (let's say MUMmer) on the instance, fetch the data (from S3 or Azure or something else), launch the software (let's say MUMmer). When the software completes, the results are pushed by to cloud storage, and the instance is killed. So in this kind of platform, the installation occurs independently for each job you launch. I don't know if Galaxy works like that -- that was my question (I am reading the 3 important Galaxy papers today !).
but if you install Ray the tool_dependency file is read and at that time it will be installed and the openmpi dependency.
So the question becomes "When Ray get installed ?" Answers may include: A) Ray is installed once by a sysadmin. B) Ray is installed automatically by Galaxy when it starts if Ray is not already installed. C) Ray is installed each time someone uses it because an instance is spawned. D) Ray is not installed because it ships with the operating system, just like ls, pwd and so on. E) the Galaxy user needs to upload a static Ray executable just like he/she uploads their data.
Cheers, Björn
-Sébastien.
Thanks ! ___________________________________________________________ The Galaxy User list should be used for the discussion of Galaxy analysis and other features on the public server at usegalaxy.org. Please keep all replies on the list by using "reply all" in your mail client. For discussion of local Galaxy instances and the Galaxy source code, please use the Galaxy Development list:
http://lists.bx.psu.edu/listinfo/galaxy-dev
To manage your subscriptions to this and other Galaxy lists, please use the interface at:
To search Galaxy mailing lists use the unified search at: