Potential bug in the function getXcmsRaw about xcms HOT 7 CLOSED

Thanks for the notification, perfect timing!. If you could hack this
into a unitTest checking your expected behaviour
(and failling if it doesn't) would be awesome,
then we could check this does not bite is in the future.
Yours, Steffen

from xcms.

Steffen,

Unfortunately, I could't create a small XcmsSet object to test the method
getXcmsRaw (maybe you could help me, any aligned XcmsSet object with two
samples would do).

I could provide you with the data I'm running it on, but I'm not sure if it
will be appropriate for RUnit test since it's a bit time consuming.

Aleksandr

On Mon, Aug 1, 2016 at 1:09 PM sneumann [email protected] wrote:

Thanks for the notification, perfect timing!. If you could hack this
into a unitTest checking your expected behaviour
(and failling if it doesn't) would be awesome,
then we could check this does not bite is in the future.
Yours, Steffen

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#44 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/AM_-kKPb0ETkMXZ_RkizSElbgemj7N82ks5qbihigaJpZM4JZsRK
.

from xcms.

I'll have a closer look at this. At first glance it looks that @asmirn1 is right. I'll add some more RUnit tests for getXcmsRaw and then we'll see.

from xcms.

OK, there is indeed a bug:

> library(xcms)
> library(faahKO)
> library(RUnit)
> data(faahko)
> attr(faahko, "filepaths") <- sapply(as.list(basename(attr(faahko, "filepaths"))),
+                                     function(x) system.file("cdf", if (length(grep("ko",x)) > 0) "KO" else  "WT" ,x, package = "faahKO"))
> xsetRaw <- updateObject(faahko)
## Subset to two files.
> xsetRaw <- xsetRaw[, 1:2]

> ## First sample is reference, i.e. no rt adjustment performed
> xs <- retcor(group(xsetRaw), method = "obiwarp", center = 1)
262 325 387 450 512 575 
center sample:  ko15 
Processing: ko16

> ## Second is corrected, first is center:
> checkIdentical(xs@rt$raw[[1]], xs@rt$corrected[[1]])
[1] TRUE
> checkTrue(!all(xs@rt$raw[[2]] == xs@rt$corrected[[2]]))
[1] TRUE 
> ## We get the raw data for the second file; this one was corrected and
> ## thus it's retention time is expected to be different from raw.
> xr2 <- getXcmsRaw(xs, sampleidx = 2, rt = "corrected")
No need to perform retention time correction, raw and corrected rt are identical for /Users/jo/R/R-3.3.1-Bioc3.4-devel/lib/R/library/faahKO/cdf/KO/ko16.CDF
> checkIdentical(xr2@scantime, xs@rt$corrected[[2]])
Error in checkIdentical(xr2@scantime, xs@rt$corrected[[2]]) : FALSE

> all(xr2@scantime == xs@rt$raw[[2]])  ## That should be FALSE!
[1] TRUE

I've wrapped that into a unit test and will fix the bug in the devel branch.

from xcms.

Dear @asmirn1 , @sneumann I fixed this bug in commit e429a1b.

Aleksandr was completely right. I've now included a RUnit test specifically for this problem and have also updated the NEWS; I didn't bump the version though, as I prefer that Steffen does that if he's OK with the changes.

from xcms.

This should appear now in xcms-1.49.5 (Due to the way the merging worked, 1.49.4 only lived
a couple of minutes).

from xcms.

I think it's OK if we close this issue now.

from xcms.