Hi Tony, Well, what I think would be best is to try and implement Bioconductor tools as plugins using BASE's existing plugin architecture. BASE can call upon a plugin (any executable program or script that resides on the server), pass along to it a plain text file containing the data and any arguments (user-definable through the built-in plugin interface) in a certain format (called the BASEfile), and execute the plugin. BASE has a jobcontroller that handles the execution of the program in the background, and when it is finished, expects back a result file(s) in a various formats which can then be imported back into the database as a resultant dataset, or perhaps an image stored for visualization. This would all be done at the level of the BioAssaySet, if you're familiar with the internal name of things in BASE. So, an important component in this case would be a handling application to convert back and forth between the BASEfile format and standard input/output formats for Bioconductor tools. This is essentially how most existing BASE plugins have been implemented (including one in R), so I think it could be a good approach for Bioconductor too. This way, it will require little to no custom modifications to get Bioconductor plugins to work with existing BASE installations. However, for plugins that also generate some kind of graphical output, additional web-based visualization tools outside of the BASE plugin architecture could be written (for instance, the hierarchical clustering plugin has a separate visualization tool). One reason we find BASE useful is that all analysis steps are stored within the system (BioAssaySets and transformations thereof are viewable to the user in a kind of hierarchical tree), so it is simple to look back and see each step, what was performed, the settings, and the result. It would be great if Bioconductor-based plugins also took advantage of this part of BASE, so that the output from for instance a Bioconductor ANOVA p-value genefilter plugin could also be stored and the resulting data run through another plugin, etc, all seamlessly from within BASE. If you and others want to tackle this in a more serious way, perhaps we should move the technical part of this discussion over to the BASE mailing list? Lao At 06:18 PM 6/11/03 -0700, A.J. Rossini wrote: >There are two general approaches that could be taken here -- one would >be server-side, to use the R PHP interface (assumed it is any good) >that was just announced on r-help. > >The other approach, which I prefer (but it's just me, perhaps), would >be to use BASE as a backing-store for computing on. > >We'd briefly talked by email before, Lao, but I've not had a chance to >follow up. I'd be interested in being in the loop on any proposals, >in case I might be able to contribute. > >best, >-tony > >-- >A.J. Rossini / rossini@u.washington.edu / rossini@scharp.org >Biomedical/Health Informatics and Biostatistics, University of Washington. >Biostatistics, HVTN/SCHARP, Fred Hutchinson Cancer Research Center. >FHCRC: 206-667-7025 (fax=4812)|Voicemail is pretty sketchy/use Email > > >Lao Saal <lhs74@columbia.edu> writes: > > > I am wondering if anyone out there has implemented any BASE plug- ins > > using Bioconductor? I'm not aware of anything yet, but see some great > > opportunities. Is anyone using BASE or interested in making > > Bioconductor tools compatible with it? > > > > For those of you not familiar with BASE, you can find more information > > at http://base.thep.lu.se/. It is basically a web-based platform for > > a biologist to manage all areas of microarray experimentation, > > including data analysis via plug-in modules. > > > > I'd like to hear your thoughts. Additionally, some of the BASE > > developers are going to be at the upcoming MGED6 meeting -- so that > > could be a good opportunity to meet and discuss the possibilities in > > person.