[Bioc-devel] RFC: read.AnnotatedDataFrame

Vincent Carey 525-2265 stvjc at channing.harvard.edu
Thu May 24 18:26:08 CEST 2007


> Hi Martin,
>
> > There have been a couple of requests for widget support, so
> > implementing rather than removing the widget option might be better
> > (though of course more difficult -- there are several different paths
> > through the code, users haven't been forthcoming about which are
> > actually important to them, and different paths have different sets of
> > bugs / design issues that make adoption to AnnotatedDataFrame more
> > challenging than you might expect).
>
> Frankly, if people can prepare a file easily with MS Excel and then call

is it a single file?  the annotated data frame should have several
components including pData, featureData, varMetadata?  (apologies
if i am misremembering the structure)

> a simple function to read it into R, I do not see much of a use case for
> a widget. Anyway, I am not in a position to implement the widgets
> functionality, but if it turns out that someone is, maybe this would be
> something for the wonderful new Qt stuff rather than tcltk?

i think the widget approach is good for a multicomponent structure.
particularly because it guides one through the components in case
they don't retain them (as i don't).  implementation in tcltk vs
qt is not a big concern.

>
> Currently, the outcome of using this option with read.AnnotatedDataFrame
> and read.AnnotatedDataFrame2 results in the same: an error. But we can
> keep it around in the function signature if it makes people happier :)
>
> > Along the same lines, the args to read.AnnotatedDataFrame were copied
> > from read.phenoData; probably there are a number of scripts out there
> > relying on these.
>
> Hmm, but read.phenoData is a different function than
> read.AnnotatedDataFrame. Is your design goal that one should be able to
> simply replace all calls to read.phenoData in existing scripts with
> calls to read.AnnotatedDataFrame? I think people that move their scripts

that seems a reasonable step -- i think read.phenoData should be
eliminated.



More information about the Bioc-devel mailing list