[BioC] Restore lost mm probes in an altcdfenv?

[lgautier at altern.org]

> Hi all,
>      Being an R novice I'm hoping someone in the community might have
> already dealt with something of this sort.
>
> I'm trying to create an altcdfenv (or analogous package) remapping the
> probes from a custom chip onto new gene sequences.  I've read the
> excellent vignette, matched the probes, made the cdfenv and all seemed
> well, but:
>
> mas5(Data)
> background correction: mas
> PM/MM correction : mas
> expression values: mas
> background correcting...Error in as.vector(data) : NA/NaN/Inf in foreign
> function call (arg 1)
>
> I looked deeper and found that this is because, as I might have realized,
> all the MM probes are set to NA by buildCdfEnv, although it happily opens
> the original cdf where the locations of all the pm/mm pairs are
> identified.  Here's one probe set in the alt.cdf:
>
>              mm
>  [1,] 219922 NA
>  [2,] 255744 NA
>  [3,] 313097 NA
>  [4,] 378979 NA
>  [5,] 214348 NA
>  [6,] 279991 NA
>  [7,] 228789 NA
>  [8,] 503799 NA
>  [9,]  78517 NA
> [10,] 253259 NA
> [11,] 367820 NA
> [12,] 407971 NA
> [13,] 280585 NA
> [14,]  68092 NA
>
> Not being very good at R, how hard would it be to recover the mm probes
> that correspond to these guys?  I'd appreciate any advice.

The MMs are set to NA because both PMs and MMs were indifferently matched
against the reference sequence; at one point not only the grouping into
probesets, but also the grouping in probe pairs was undone before
matching.

MMs being thought of little use at the time, it was left this way... but
does not have to be that way (and nowadays MMs are used by methods such as
gcRMA).

That would not be particularly hard, but will currently require a bit
of R programming. The place to modify is buildCdfEnv.matchprobes, while
looking what the function matchprobes (package matchprobes) is doing.
I'll try have that in for the next release.


Laurent


> Thanks,
> Mike
>
>
>
>
>       ____________________________________________________________________________________
> Be a better friend, newshound, and
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor
>