[Bioc-devel] biocViews preprocessing term

Sure, I can do that. Just need to figure out if there is an easy way
to parse through the whole repository, I hope that Patrick can help
with that. However, I still don't see how one would specify a term
like flow cytometry -> preprocessing in the DESCRIPTION file. Or do yo
mean unique node names by syntactic atoms? So far, only the leaf node
seems to be used to assign a package to a category.

Florian


On 03.09.2008, at 03:42, Vincent Carey 525-2265 wrote:

Thanks for carrying on this thread, Patrick,
looks like I got distracted.

Now I see how the vocabulary is maintained.
biocViews/inst/dot has a nice human-readable rendering
of the vocabulary in biocViewsVocab.dot

Each node reference in the dot is a topic, and any
node in that dot file can be used as a biocViews
field value in the package.

Florian, would you like to make the changes to
biocViews/inst/dot that reflect your desired vocabulary
modifications, rebuild the vocabulary using biocViews/updateVocab.sh,
and then commit the modified biocViews package (with a bump to the
version tag in DESCRIPTION)?  It would be nice to have some
mechanism of testing that the build system will survive such
a change (perhaps testing some validity conditions on the new
graph, such as 1 connected component, all nodes are syntactic
atoms, checking the correspondence between biocViews fields
present in the repository and the nodes in the graph).

I will do the changes mentioned in Florian's e-mail and commit
if people want to funnel this through me ...

---
Vince Carey, PhD
Assoc. Prof Med (Biostatistics)
Harvard Medical School
Channing Laboratory - ph 6175252265 fa 6177311541
181 Longwood Ave Boston MA 02115 USA
stvjc at channing.harvard.edu

On Wed, 3 Sep 2008, Patrick Aboyoun wrote:

On the second point, the build system uses the ontology in
biocViews and
will adapt accordingly when biocViews is changed. Any vocabulary term
that is not recognized by biocViews will be ignored and an internal
build warning will be issued. Periodically I check the ignored
terms to
see if this is due to misspellings (which I correct), but in some
situations packages include categories that don't exist. If changes
are
made to the ontology, package authors can modify their biocViews
DESCRIPTION field prior to changes being made to biocViews without
any
adverse build outcome.


Patrick



Vincent Carey 525-2265 wrote:

yes in specific, and although you did not ask it does
appear that we need some renovation of the biocViews
vocabulary.

alas i am not really clear on how modifications to the
vocabulary impact the build system.  i will take a look
at this and make a proposal momentarily.

---
Vince Carey, PhD
Assoc. Prof Med (Biostatistics)
Harvard Medical School
Channing Laboratory - ph 6175252265 fa 6177311541
181 Longwood Ave Boston MA 02115 USA
stvjc at channing.harvard.edu

On Tue, 2 Sep 2008, Florian Hahne wrote:

Hi Vince,
the cellHTS and cellHTS2 package provide (mostly) preprocessing
functionality for cell-based assays, and assigning the biocViews
term
"Preprocessing" seems to be reasonable. However, in the ontology
this is
defined as "Microarray -> Preprocessing;", and that's not really
what we
want. Shouldn't there be a general "Preprocessing" term
independent of
micro arrays. I guess this will also become interesting for many
of the
other emerging new technologies like deep sequencing. With the
current
design, it seems this is not possible, since terms have to be
unambiguous. Is there any reasonable way of dealing with this
problem
other than adding a new term like "General Preprocessing"? And
wouldn't
it be useful to allow for terms like "Technology -> FlowCytometry
->Preprocessing;"

Florian

--
Florian Hahne, PhD
Computational Biology Program
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N, M2-B876
PO Box 19024
Seattle, Washington 98109-1024
206-667-3148
fhahne at fhcrc.org