[Bioc-devel] Base class for interaction data - expressions of interest

In our discussions of the multiassay object/analysis problem,

https://github.com/vjcitn/biocMultiAssay

Kasper Hansen has suggested elaborations of the Hits API (see for example
sec. 4 of


http://bioconductor.org/packages/release/bioc/vignettes/GenomicRanges/inst/doc/GenomicRangesIntroduction.pdf

,

more fully described in man pages in S4Vectors


http://bioconductor.org/packages/release/bioc/manuals/S4Vectors/man/S4Vectors.pdf
)

to manage work on general feature-feature relationships.  At present, the
Hits structures/methods
use a pair of integer vectors to collate elements of two objects.  Two
general problems that probably need more work are a) classes for feature
annotation elements and b) extensions to Hits that support R idioms for
subsetting collated objects at various points on an analysis path.

On Sun, Nov 8, 2015 at 8:05 AM, Laurent Gatto <lg390 at cam.ac.uk> wrote:

On  7 November 2015 19:58, Ludwig Geistlinger wrote:

I agree with Martin, I would love to see something like that.
Especially if this would not be restricted to chromatin interactions,

but

also allows to represent protein-protein,

transcriptionFactor-targetGene,

miRNA-mRNA etc (e.g. via suitably tailored subclasses).

I would be delighted to help out in PPI (or Protein-anything
interactions) infrastructure development, if this is in scope.

Best wishes,

Laurent

This might nicely work together with bringing in existing regulatory
networks via AnnotationHub ...

If help is needed, I'm happy to contribute ...

Best,
Ludwig

Just to say that this is a great idea. If this starts as a github

package

(or in svn, we can create a location for you if you'd like) I and

others

would I am sure be happy to try to provide any guidance / insight. The
main design principles are probably to reuse as much as possible from
existing classes, especially the S4Vectors / GRanges world, and to
integrate metadata as appropriate (like SummarizedExepriment, for
instance).

Martin

Aaron Lun

[alun at wehi.edu.au]
Sent: Thursday, November 05, 2015 12:27 PM
To: bioc-devel at r-project.org
Subject: Re: [Bioc-devel] Base class for interaction data -

expressions of

     interest

There's a growing number of Bioconductor packages dealing with
interaction data; diffHic, GenomicInteractions, HiTC, to name a few

(and

probably more in the future). Each of these packages defines its own
class to store interaction data - DIList for diffHic,
GenomicInteractions for GenomicInteractions, and HTClist for HiTC.

These classes seem to share a lot of features, which suggests that they
can be (easily?) replaced with a common class. This would have two
advantages - one, developers of new and existing packages don't have to
continually write and maintain new classes; and two, it provides users
with a consistent user experience across the relevant packages.

My question is, does anybody have anything in the pipeline with respect
to a base package for an interaction class? If not, I'm planning to put
something together for the next BioC release. To this end, I'd welcome
any ideas/input/code; the aim is to make a drop-in replacement (insofar
as that's possible) for the existing classes in each package.

Cheers,

Aaron

employee or

agent responsible for the delivery of this message to the intended
recipient(s), you are hereby notified that any disclosure, copying,
distribution, or use of this email message is prohibited.  If you have
received this message in error, please notify the sender immediately by
e-mail and delete this email message from your computer. Thank you.

--
Laurent Gatto | @lgatt0
http://cpu.sysbiol.cam.ac.uk/
http://lgatto.github.io/