Thanks Ross, the -v option is definitely a good way to go that I had
not considered. For anyone else following this thread, the tool
dependency environment system was just commented on and documented by
See the following links:
On Tue, Jan 24, 2012 at 8:27 PM, Ross <ross.lazarus(a)gmail.com> wrote:
Without knowing exactly what 'a cluster with a similar setup to
CloudMan' really means, if you're using SGE, passing the galaxy
process's path to the job is usually accomplished by setting the -V
flag in sge_request
eg on an ubuntu 11.10 machine with 'default' as the queue
(vgalaxy)galaxy@iaas1:$ cat /var/lib/gridengine/default/common/sge_request
# system wide defaults
works for me.
On Wed, Jan 25, 2012 at 12:12 PM, Andrew Warren <anwarren(a)vbi.vt.edu> wrote:
> Thanks for the replies. One extra element I failed to properly describe is
> that we are running on a cluster with a similar setup to CloudMan. This
> means we have a galaxyTools/tools directory with folders for each tool, and
> subfolders for each version with a "default" symlink to the version
> currently in use. Each subfolder has a "env.sh" script which is added by
> dependency manager to the qsub script to be processed at runtime . I was
> thinking that one of the benefits of using the "requirements" tag is that
> would/could allow for tool and version specific dependency chains. For
> instance if a version of tophat only runs with a certain version of bowtie
> but you want to make the newest beta version of bowtie also available to run
> Right now the (sort of problem) I am having is that in order for the
> necessary PATH information to be transmitted to the compute nodes I have to
> have bowtie and samtools listed as requirements in the tool wrapper. The
> PATH variable of the galaxy user doesn't transmit to the compute nodes
> (despite having it set before launching the main instance of galaxy). With
> our current setup I have seen this with several different tools that call
> other tools and so I am wondering if my cluster/cloud setup has gone wrong
> somewhere. Is there a "right" way or place to specify the runtime PATH so
> that it will be transmitted to the compute nodes? Just trying to figure this
> out in terms of "best practices" since my current setup seems to require
> modification of the default wrappers to transmit PATH information to the
> compute nodes.
> Thanks again,
> On Sun, Jan 22, 2012 at 8:00 AM, Hans-Rudolf Hotz <hrh(a)fmi.ch> wrote:
>> On 01/22/2012 01:41 AM, Anthonius deBoer wrote:
>>> All tools need to be in the path of the user running galaxy.
>> but can be simply added to the path by adding them to the 'run.sh'
>> Regards, Hans
>>> Thon de Boer, Ph.D
>>> Bioinformatics Guru
>>> T: +1.650.799.6839 | E-mail: thondeboer(a)me.com
>>> "IMPORTANT NOTICE: This email message is legally privileged,
>>> and is for the use of the individual or entity to whom it is addressed. If
>>> you have received this email message by error, please immediately notify us
>>> by email and delete the message. Thank you."
>>> On Jan 21, 2012, at 3:50 PM, Andrew Warren<anwarren(a)vbi.vt.edu>
>>>> We recently transitioned from a CloudMan instance of galaxy to our own
>>>> cluster and started having problems with calls to tools from within
>>>> other tools. For example when Tophat calls bowtie-inspect its not
>>>> finding the executable. To fix this I listed bowtie in the
>>>> requirements section of the tophat wrapper like so:
>>>> <tool id="tophat" name="Tophat for Illumina"
>>>> <description>Find splice junctions using RNA-seq
>>>> <version_command>tophat --version</version_command>
>>>> Now I am wondering, is it generally expected that all tools used by
>>>> galaxy will have their executables on the user galaxy's PATH? Is
>>>> above a good solution? Or is there something else likely amiss with
>>>> our galaxy setup? I think we recently pulled updates for some major
>>>> tool_shed release but I haven't been able to determine if any of
>>>> tools listed above were affected by that.
>>>> Wish I were in Český Krumlov asking this question. Missed the
>>>> registration deadline...doh.
>>>> Andrew Warren
>>>> Please keep all replies on the list by using "reply all"
>>>> in your mail client. To manage your subscriptions to this
>>>> and other Galaxy lists, please use the interface at:
>>> Please keep all replies on the list by using "reply all"
>>> in your mail client. To manage your subscriptions to this
>>> and other Galaxy lists, please use the interface at:
> Please keep all replies on the list by using "reply all"
> in your mail client. To manage your subscriptions to this
> and other Galaxy lists, please use the interface at:
Ross Lazarus MBBS MPH;
Associate Professor, Harvard Medical School;
Head, Medical Bioinformatics, BakerIDI; Tel: +61 385321444;