I also discovered another apparent bug later in the same function. The second to last line of makeVectorsAffyBatch is

vers <- ifelse(!is.null(cdfname), as.character(packageVersion(cdfname)), "")

If cdfname is NULL, this line will throw an error because the second argument to ifelse will have length zero and "ifelse" does NOT do lazy evaluation.

On 07/06/2015 12:21 PM, Ryan C. Thompson wrote:
Hello,

I just encountered a bug in frmaTools that makes it impossible to use on certain array platforms. The following lines in makeVectorsAffyBatch fail on an AffyBatch object on the hthgu133pluspm platform:

    pms <- pm(object)
    pns <- probeNames(object)
    pmi <- unlist(pmindex(object))
    if (!identical(as.character(pmi), rownames(pms)))
        stop("Mismatch between pmindex and rownames of pms")

I isolated the problem to five probes:

> i <- which(as.character(pmi) != rownames(pms))
> pmi[i]
1564498_PM_at9 205398_PM_s_at8 217695_PM_x_at7 223446_PM_s_at7 237802_PM_at3
          3e+05           5e+05           2e+05 1e+05 4e+05
> rownames(pms)[i]
[1] "300000" "500000" "200000" "100000" "400000"
> as.character(pmi)[i]
[1] "3e+05" "5e+05" "2e+05" "1e+05" "4e+05"

As you can see, the problem is that as.character will happily use scientific notation when it feels like it, which then fails a test for string equality. I believe the solution is to replace that test with:

all(sprintf("%i", pmi) == rownames(pms))

-Ryan

_______________________________________________
Bioc-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel

Reply via email to