[BioC] IlluminaHumanMethylation450k.db: Missing probes in IlluminaHumanMethylation450kPROBELOCATION function?

[Simone]

Dear Tim,

Thanks a lot for your comprehensive reply. The code you posted seems
quite complicated, I'll have a closer look on it tomorrow. I am just
writing already now to tell you that I made a mistake in my previous
posting: I did not mean IlluminaHumanMethylation450kPATH2PROBE, but
IlluminaHumanMethylation450kPROBELOCATION. I mixed it up with some
tests I did regarding toggleProbes(). Sorry!

However, I get the same error for
IlluminaHumanMethylation450kPROBELOCATION (which is probably the
reason why I finally mixed it up):

 x <- toggleProbes(IlluminaHumanMethylation450kPROBELOCATION, "all")
Error in (function (classes, fdef, mtable)  :
  unable to find an inherited method for function ‘toggleProbes’ for
signature ‘"AnnDbBimap"’

Best wishes,
Simone

On Tue, Jul 2, 2013 at 7:25 PM, Tim Triche, Jr. <tim.triche at gmail.com> wrote:
> re: IlluminaHumanMethylation450kPATH2PROBE
>
> This one is new to me.  I suspect that not all probes have a pathway mapped
> to them, but I will have to investigate further.  If it's a KEGG issue, it
> may not be possible to resolve this due to KEGG's deprecation.
>
>
>
> On Tue, Jul 2, 2013 at 2:14 AM, Simone <enomis.bioc at gmail.com> wrote:
>>
>> Dear Tim,
>>
>> > You will need to use toggleProbes() to get all probe locations (this is
>> > an
>> > undesirable side effect of the db0 package infrastructure, which is
>> > geared
>> > towards expression arrays where probes that map to multiple accessions
>> > are a
>> > "bad thing", and that is the reason I deprecated the 27k.db and 450k.db
>> > packages).
>>
>> I understand. However, also with toggleProbes() I have got a problem.
>> I tried for example:
>>
>> > x <- toggleProbes(IlluminaHumanMethylation450kPATH2PROBE, "all")
>>
>> Error in (function (classes, fdef, mtable)  :
>>   unable to find an inherited method for function ‘toggleProbes’ for
>> signature ‘"AnnDbBimap"’
>>
>> But for example the following works
>>
>> > x <- toggleProbes(IlluminaHumanMethylation450kENSEMBL2PROBE, "all")
>>
>> However, I would need the location information, so the column
>> "transcript_location" (e.g. 5'UTR, TSS200, Body, etc.) and it seems
>> that I cannot get it like this?
>>
>> > The FDb.InfiniumMethylation packages have the locations of the targeted
>> > CpGs
>> > (and CpHs, and SNPs) for the corresponding build (just for the record,
>> > the
>> > genome build targeted by 450k.db is hg19/GRch37), but they are returned
>> > using get450k() or get27k() as a GRanges object.
>>
>> Currently, this doesn't work for me neither (see my previous post).
>> Probably I'm doing something wrong ...
>>
>> > Essentially, the mistake I made in compiling the 27k.db and 450k.db
>> > packages
>> > was to use an expression-centric (or gene- and transcript-centric)
>> > infrastructure to represent probes targeting individual loci.  It's more
>> > useful to think of the 450k array, in particular, as a SNP array, with
>> > functions for a given target locus inferred after discovery.  It's also
>> > more
>> > useful to think of the chip as a SNP array when considering
>> > differentially
>> > methylated regions, which could be loosely regarded as the methylation
>> > equivalent of copy number variations.  Last but not least, you can in
>> > fact
>> > use the array to estimate copy number along the genome, as for
>> > determining
>> > sex from chrX copy number.
>>
>> Thank you for your hints and explanations!
>> But at the moment I am still stucked at a much simpler step, I would
>> just need an easy way to distinguish probes located in promoters from
>> those located in gene bodies (and to discard those which are annotated
>> to both).
>>
>> Best wishes,
>> Simone
>
>
>
>
> --
> A model is a lie that helps you see the truth.
>
> Howard Skipper