[Bioc-devel] RFC for a container for ChIP-chip data

I am working on an experimental data package for the Harbison
(PMID 15343339) yeast regulatory code paper.

the assay data can be looked at either in a gene-centric way (e.g.,
reporters are genes; most convenient, but some ambiguity) or more
literally using the intergenic region, for which we have sequence
information.  the intergenic regions are the things that are
actually spotted.

my plan at the moment -- define a ChIPset class that extends
eSet; the AssayData will probably have accessor brats (for
binding ratios, not exprs), and i will try to set up featureData
to have some information on associated intergenic region for
a gene-centric representation of the assay.  The phenoData will
give information on the TFs that were chipped for each AssayData
column

any concerns with this plan?  any other formats for proprietary
ChIP-chip data that I don't know about but should?

column

any concerns with this plan?  any other formats for proprietary
ChIP-chip data that I don't know about but should?

Vince,

This looks good for non-tiling Chip-chip data, but not for tiling or
promoter-tiling chips which are rapidly becoming the standard (?) for
performing chip-chip analysis.  I would vote for making some distinction
between gene-based chip-chip data and tiling/promoter-based chip, as the
analysis methods and featureData are quite distinct from PCR-product based
arrays.  This is probably just a naming thing, but I think ChIPset implies
different things to different people.  Or, if you go with ChIPset, I guess a
TilingChIPset could be created?  I'm not proposing a solution, but just
pointing out that to many people, ChIP-chip data will imply tiling-array data
with genomic locations, etc. and to others it will not.

Sean

I am working on an experimental data package for the Harbison
(PMID 15343339) yeast regulatory code paper.

the assay data can be looked at either in a gene-centric way (e.g.,
reporters are genes; most convenient, but some ambiguity) or more
literally using the intergenic region, for which we have sequence
information.  the intergenic regions are the things that are
actually spotted.

my plan at the moment -- define a ChIPset class that extends
eSet; the AssayData will probably have accessor brats (for
binding ratios, not exprs), and i will try to set up featureData
to have some information on associated intergenic region for
a gene-centric representation of the assay.  The phenoData will
give information on the TFs that were chipped for each AssayData
column

any concerns with this plan?  any other formats for proprietary
ChIP-chip data that I don't know about but should?

Rafa and Benilton directed me to their TilingFeatureSet class which
will cover the tiling case.  As you note the Harbison example is
different and I do not want to steal the ChIPset name for this
possibly outmoded approach.  maybe nonTilingChIPset?  

I don't know how many exemplars we may want to have of this kind of
data, so I don't want to sweat it too much.

If no one anticipates multiple exemplars of this kind of data, we
could just use ExpressionSet and live with the terminologic abuse.

 Hi all,

Does anybody have experience with using bibtex in vignettes? In the
latest version of my package I switched to bibtex for managing the
citations in the vignette of my package, including a .bib file in
inst/doc. This seemed to work fine when I tested it myself. But in the
final release vignette I see all my references replaced by question
marks :-(.

I don't know how many exemplars we may want to have of this kind of
data, so I don't want to sweat it too much.

If no one anticipates multiple exemplars of this kind of data, we
could just use ExpressionSet and live with the terminologic abuse.

If the only benefit of a new class is new names and if you will want
to apply the same computations, then you really are just making work.
Make a new class _if_ the existing methods (thinking machine learning,
etc) that operate on exprs() are not going to make sense and need
customization.

 Hi all,

Does anybody have experience with using bibtex in vignettes? In the
latest version of my package I switched to bibtex for managing the
citations in the vignette of my package, including a .bib file in
inst/doc. This seemed to work fine when I tested it myself. But in the
final release vignette I see all my references replaced by question
marks :-(.

i think it may be a reasonable request to add bibtex processing
into the build, but i don't know how easy it is to implement.
in the mean time, couldn't you add the contents of the .bbl file
to your Rnw as a literal included bibliography?  then as long
as latex is run twice i think you should have the citations
resolved?

then the build would (i think) need to run latex twice.

Vincent Carey 525-2265 <stvjc at channing.harvard.edu> writes:

 Hi all,

Does anybody have experience with using bibtex in vignettes? In the
latest version of my package I switched to bibtex for managing the
citations in the vignette of my package, including a .bib file in
inst/doc. This seemed to work fine when I tested it myself. But in the
final release vignette I see all my references replaced by question
marks :-(.

i think it may be a reasonable request to add bibtex processing
into the build, but i don't know how easy it is to implement.
in the mean time, couldn't you add the contents of the .bbl file
to your Rnw as a literal included bibliography?  then as long
as latex is run twice i think you should have the citations
resolved?

then the build would (i think) need to run latex twice.

Uh, I think this Just Works.  There are many existing packages that
have a .bib file inside inst/doc and the references seem to be present
in the vignettes.  Looks at GOstats and affycoretools, for example.

+ seth

Just to clarify, I think this is occurring with the Windows build only
-- look in the .zip file for globaltest. Martin

Just to clarify, I think this is occurring with the Windows build only
-- look in the .zip file for globaltest. Martin
    

And it also runs fine without question marks on my own windows machine.
But obviously I do not have the same LaTeX installation. I'll try to
imitate the precise bibtex settings of other vignettes that use bibtex.
Let's see if that helps...

Jelle