[Bioc-devel] deprecation of keepSeqlevels and renameSeqlevels

On Mon, May 20, 2013 at 3:11 PM, Martin Morgan <mtmorgan at fhcrc.org> wrote:

Hi Michael --


On 5/20/2013 5:56 AM, Michael Lawrence wrote:

While it's cool that seqlevels<- has become so flexible, I still claim
that
rename/keep/drop would be a lot easier to read, because they describe the
high-level operation, and there's no reason to deprecate them. They're
also
easier to remember. For example, for dropping with seqlevels<-, the user
needs
to remember that "force" is necessary to drop. Much easier to just say
"dropSeqlevels(), please". And reimplementing keepSeqlevels is still too
complicated. Is it such a maintenance burden to have these simple
wrappers sit
on top of the low-level manipulators?

Another suggestion: renameSeqlevels should not require a named vector (in
cases
where it is unnamed, require that it is parallel to the existing
seqlevels, as
with seqlevels<-).

I didn't really indicate what drove my desire to see keepSeqlevels
deprecated. keepSeqlevels, seqlevels<-, and isActiveSeq were developed more
or less independently, and have different contracts. The different
contracts are confusing to the user, as is creating a usable help system
when there are multiple end points for what is a common operation. The help
pages of each were inadequate in different ways. And because the code was
developed independently, support for different objects was inconsistent. So
actually, yes, the maintenance (and use) burden for the previous state of
affairs was substantial.

On the other hand, I agree that keepSeqlevels is convenient as a simple
wrapper around seqlevels<-, in the same way that setNames and names<- are
both useful.

So we could iterate to

   keepSeqlevels <-
       function(x, value, ...)
   {
       seqlevels(x, force=TRUE) <- value
       x
   }

but I'd be less enthusiastic about maintaining the original contract of
keepSeqlevels, where keepSeqlevels(gr1, gr2), would have worked for two
GRanges objects.

Well, I wasn't even aware of that feature, so you've made your point about
the documentation ;) Sounds like a good compromise to me.

Thanks for understanding,
Michael

Martin

Michael






On Sat, May 18, 2013 at 6:00 PM, Martin Morgan <mtmorgan at fhcrc.org
<mailto:mtmorgan at fhcrc.org>> wrote:

     On 05/18/2013 05:42 PM, Martin Morgan wrote:

         Some of the most common operations are straight-forward, e.g.,

               > gr = GRanges(paste0("chr", c(1:22, "X", "Y")), IRanges(1,

100))

               > seqlevels(gr) = sub("chr", "ch", seqlevels(gr))

     To get a more comprehensive example I should have followed my own
advice and
     grabbed from the help page!

           ## Rename:
           seqlevels(txdb) <- sub("chr", "", seqlevels(txdb))
           seqlevels(txdb)

           seqlevels(txdb) <- paste0("CH", seqlevels(txdb))
           seqlevels(txdb)

           seqlevels(txdb)[seqlevels(__**txdb) == "CHM"] <- "M"

           seqlevels(txdb)


     --
     Computational Biology / Fred Hutchinson Cancer Research Center
     1100 Fairview Ave. N.
     PO Box 19024 Seattle, WA 98109

     Location: Arnold Building M1 B861
     Phone: (206) 667-2793 <tel:%28206%29%20667-2793>

--
Dr. Martin Morgan, PhD

Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.
PO Box 19024 Seattle, WA 98109

	[[alternative HTML version deleted]]