[Bioc-devel] ANN: BioC Developer's Meeting/agenda/discussion

James Wettenhall wettenhall at wehi.EDU.AU
Mon Jul 18 08:16:14 CEST 2005


Hi,

I hope you don't mind a less senior (non-BioC-core) member commenting on
this discussion...

I think the issue of whether one package should do one thing or multiple
things is related to the issue of whether BioC is for researchers or for
users who expect commercial-style software quality assurance.  Here at the
WEHI, we do complete microarray analyses for many biologists, mainly using
the limma package, and we consider opportunities to enhance all steps of
the analysis in a research sense.  But I think limma also has pretty good
software quality assurance thanks to Gordon and maybe also because it has
so many users to give feedback.  It is certainly convenient for us to have
only one (or a small number of) packages to install when we do microarray
analysis at WEHI.  If we regularly used 10 different BioC packages then I
would feel less confident that they would all work together with each
other, because it seems to me that leading up to each "stable" release of
BioC, there are still quite a lot of major changes relative to
testing/quality-assurance, for example some changes in hexbin for BioC 1.6
Release broke the arrayQuality package for BioC 1.6 Release.  In addition
to my nervousness about a larger number of packages (and their
dependencies) working together, the other reason why it's nice to have
overlapping features in BioC packages is that it allows some independent
competitive research of these overlapping features - I imagine that many
BioC developers would like some flexibility to do research and have the
results implemented in a package they can quickly update and test, rather
than waiting for someone in a different time zone to consider adding and
testing their new features in the package, and possibly ending up with a
package that depends on a development version of R.

Unfortunately, I won't be at the meeting in Seattle.  I hope it goes well!

Best wishes,
James

P.S. A colleague of mine here at WEHI, Keith Satterley, will be taking
over development of limmaGUI and affylmGUI soon.



More information about the Bioc-devel mailing list