[Bioc-devel] RFC: xy2i and i2xy in *cdf packages

I found an interesting "trick" adopted within the package mtachprobes
to solve this problem...
i.e. obtaining chip dimensions information from current versions of
CDF packages.

Within the package "matchprobes" there's a function named
".lgExtraParanoia"

If you read this function code you can find the following lines:
#############
tab = table(mm1 - pm1)
sizex = as.numeric(names(tab))[max(tab) == tab]
pm2 = pt$y * sizex + pt$x + 1
#############

"mm1" is a vector containing MM probes indexes (obtained from CDF package)
"pm1" is a vector containing PM probes indexes (obtained from CDF package)
Then chip dimension (i.e. "sizex") is obtained as the maximum difference between
MM and PM probes indexes. This is based on the assumption that PM and
MM probe are usually next to each other on the chip
at same x coordinate and at adjacent y coordinates.
The last line is equal to function "xy2i()" since "pt" is a data.frame
containing probe sequences information
and columns x and y (i.e. vectors "pt$x" and "pt$y") contain xy probes
coordinates



However I agree with the fact that this could not be the final solution
and that having dim environment accessible within CDF packages is a
better solution.

Best,
Francesco

Date: Wed, 11 Apr 2007 16:53:55 -0400
From: "James W. MacDonald" <jmacdon at med.umich.edu>
Subject: Re: [Bioc-devel] RFC: xy2i and i2xy in *cdf packages
To: Seth Falcon <sfalcon at fhcrc.org>
Cc: bioc-devel at stat.math.ethz.ch
Message-ID: <461D4AE3.2070501 at med.umich.edu>
Content-Type: text/plain;  charset="utf-8";  format=flowed

Seth Falcon wrote:

Wolfgang Huber <huber at ebi.ac.uk> writes:

Hi,

Rather than make up these new dimension objects in the CDF package,
why not teach the affy:indices2xy function to reach into the
appropriate CDF package and grab the *unexported*, existing i2xy
function in order to do its thing, and the affy:xy2indices to grab the
*unexported* xy2i function? Seems like the least intrusive way to do
things, and more flexible.

I don't think we should change the visibility of the xy2i functions
without going through the deprecation cycle.

But in general, I like your suggestion.  Teaching affy to grab the
appropriate function seems sensible.  Personally, I think it is fine
to have these functions exported.  You can still get the right one
when using affy::xy2indices.  Symbol masking is a fact of R.

But having the dimensions accessible in the CDF package also makes
sense to me and is easy so I'm still in favor of adding hgu95av2dim.
And the xy2i and i2xy functions should use it...

The functions in affy don't care what happens with xy2i and i2xy since
they don't use them (they get dimensions directly from the AffyBatch).
The only functions that care are those that don't have an AffyBatch to
query.

So, unless there are any objections*, I will add the little dim env to
the cdf packages any any functions that need this info (and don't have
an AffyBatch to query) will have to learn to use it.

*By objections, I mean 'This will completely break my code and destroy
my life'. The cdf packages are nearing the end of their shelf lives,
soon to be replaced with SQLite packages, so arguments about stylistic
differences and 'How things should be done' are counterproductive,
especially given how close we are to release. Also, remember that xy2i
and i2xy are only being _deprecated_. Anybody who uses these functions
has > 6 months from now to make changes.

Best,

Jim



--
James W. MacDonald, M.S.
Biostatistician
Affymetrix and cDNA Microarray Core
University of Michigan Cancer Center
1500 E. Medical Center Drive
7410 CCGC
Ann Arbor MI 48109
734-647-5623


**********************************************************
Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues.



------------------------------