Thanks Seth, The more I thought about this the more I came to the same conclusions. Basically we are trying to most efficiently utilize our hardware resources for our R/Bioconductor users. My current plan is to deploy R/Bioconductor on multiple Citrix servers and do load balancing via the Citrix software. This will at least make sure that each new R/Bioconductor user connection will be launched on the least busy server. Crude but it will probably server our needs (plus it will be much easier than administering a cluster). Thanks again. Pat -----Original Message----- From: Seth Falcon [mailto:sfalcon@fhcrc.org] Sent: Friday, April 20, 2007 7:11 AM To: Patrick Durham Cc: bioconductor at stat.math.ethz.ch Subject: Re: [BioC] Bioconductor on a Rocks cluster Hi Patrick, "Patrick Durham" <patrickd at="" crab.org=""> writes: > Hello, > > I'm trying to find information about running Bioconductor in a > clustered environment (specifically Rocks). I've searched Google and > the mailing list and I haven't found much at all. I am interested in > hearing from anyone who is successfully running Bioconductor on a > cluster. My first question is whether or not it will actually run on > a cluster. If so I'd would appreciate any links or documents you > could forward to me on a clustered implementation. I'm not sure Bioconductor is enough of a coehesive whole ask a question like "does Bioconductor run on a cluster"? But in general, R and Bioconductor packages will run on a cluster -- meaning you should be able to launch an R process that makes use of Bioconductor packages on a given cluster node. How much this helps you will depend on what you are trying to do. There are a few packages that help you do parallel computations (a good example is MLInterfaces, I believe). But at present, most BioC packages won't parallelize without some custom coding effort. + seth -- Seth Falcon | Computational Biology | Fred Hutchinson Cancer Research Center http://bioconductor.org