Entering edit mode
Joseph Barry
▴
160
@joseph-barry-5000
Last seen 8.1 years ago
Dana-Farber Cancer Institute, Boston, U…
Dear Juliane,
I have cc'd this email to the bioconductor mailing list as it may be
of general interest.
Thanks for isolating this issue in a neat example. It really makes
things easier to debug! As a result I have recreated the error locally
and been able to resolve it. In a nutshell, the error occurs due to an
ordering problem of the measurementNames argument to buildCellHTS2().
order(c('Channel 1', 'Channel 2', 'Channel 10'))
makes it clear why. channelNames() enforces precisely this ordering of
measurementNames later in the buildCellHTS2() function, which results
in the error that you report.
I have introduced an extra ordering step into the buildHTS2() function
in cellHTS2-devel to resolve the issue. If you checkout and install
the latest for cellHTS2-devel (see http://wiki.fhcrc.org/bioc/SvnHowTo
for details), this should resolve your problem.
Best wishes,
Joseph Barry
On Jun 4, 2012, at 9:44 AM, Siebourg Juliane wrote:
> Dear Joseph Barray,
>
> I am using your R Bioconductor package cellHTS2 and came across
something I do not understand (it might be a bug?).
> I have screening data with many dimensions (channels). When I create
cellHTS objects with the function 'buildCellHTS2(xd)' I noticed the
following: Whenever there are more than 9 different Channels, they
somehow get miss-named or shuffled in the cellHTS object.
> Here is a small example showing the issue:
>
>
> wells = sprintf("%s%02d", rep(LETTERS[1:4], each=6), 1:6)
>
> xd = expand.grid(well=wells,plate=1:3, replicate=1:2)
>
> nc<-10 #number of channels
>
> data<-matrix(rep(1:nc,nrow(xd)),nrow=nrow(xd),byrow=TRUE)
>
> colnames(data)<-paste('Channel',1:nc,sep='_')
>
> xd<-cbind(xd,data)
>
> x = buildCellHTS2(xd)
>
>
> head(xd)
>
> head(Data(x)[,1,])
>
> If you change nc to something smaller than 10 everything runs fine.
Am I doing something wrong?
> I use the buildCellHTS2 function since I get my data already in a
csv format.
>
> Thanks in advance for your help,
> Juliane Siebourg