Analyzing collaboratively HTS data -- howto??

[lcollado]

Hello,

A few of us in my lab want to analyze some HTS data; Illumina GAIIx in this case. Before we start I would love to hear some advices on how to organize yourself.

Our idea is to build an R "data package" while using Subversion (SVN) to version-control our scripts. We want to have a well documented "run-all" script (a vignette file), individual scripts that solve a small part of the analysis, and use the R package framework to document our functions and data tables. The big data files would be excluded from the package. By using an R package to join all the table results (kind of analogous to creating a zip with all the result files) the biologists on the lab will easily load the results on their computers -- yup, we've been teaching them the R basics.

While the project could make it to the public eventually, it will be "lab-eyes-only" for a while. Now, taking this into account I'm puzzled as to which SVN hosting service to use. I think that we are not really making "open source software". The other option would be to use a single account on a server and use SVN "locally" (check this). Sadly using a local server as the SVN repository is complicated for us as the IT people are very restrictive -- they've had bad luck with exterior attacks.

Any tips from your experience are more than welcome. I found this paper to be quite useful.

Thank you and greetings,
Leonardo


PS I'll be asking on the bioc-sig-sequencing (R) mailing list as well.

[lcollado]

Err, just to make it clearer. Its not that we'll do everything with R, but we'll keep all the results in R. We might "call" an external aligner from R, parse the results outside of R and then just load the final results in R.

Meaning, we'll be using R mainly instead of shell. As I see it, the advantage is on the documentation side because R will "force" us to document every table result. Also, it could make easier the process of joining results from different "modules". Plus, I like R plots

Leo