That does clear things up. I actually determined that this was the case by looking at the columns in the BeadLevel data (no Grn.bc), but found the docs confusing. Incidentally, after kludging the C source files of the beadarray package to force the parallel code, it's spectacularly fast at reading the data (for some reason, the -fopenmp flag in gcc 4.2.1 on OS X doesn't properly respond to ifdefined _OPENMP, so I removed the ifdefs altogether and used the multithreaded code lines directly). Not sure why it wouldn't compile multithreaded in the first place, but someone more familiar with the OS X compiler system could probably get it to work. It's nice to use the Mac Pro's cores. Thanks, Anand On Mon, Jan 10, 2011 at 5:34 AM, Mark Dunning <mark.dunning at="" gmail.com=""> wrote: > Hi Anand, > > I think I understand the source of your confusion. There are two types > of correction applied to BeadArray data; one at the bead-level and one > at the summarized level. > > The bead-level text files created by Illumina's scanning software > report the intensity of each bead on the array. These have been > adjusted by subtracting a local background measure calculated using > nearby pixels on the image. For convenience, the readIllumina function > in beadarray will read these values, as these are the same values that > are analysed by Illumina's BeadStudio/GenomeStudio software. We find > them adequate for most purposes, although a more thorough analysis is > possible as described in the imageProcessing vignette. If you have > data exported from BeadStudio they will almost certainly have been > adjusted in this way. > > However, the backgroundNormalise method (that is now removed) referred > to in previous emails occurs on a per bead-type basis once data have > been summarized. It used to be the default normalization method in > BeadStudio. The quantities subtracted are negative controls > (bead-types with no target in the genome). As I mention in previous > posts there are viable alternatives to this method. The neqc method > you are interested in requires summarized data and (I suspect) > developed with GenomeStudio data in mind. Therefore it is Ok to use > data that have been read using readIllumina with this method. > > Hope this helps. Regards, > > Mark > > On Thu, Jan 6, 2011 at 8:37 PM, Anand Patel <acpatel at="" gmail.com=""> wrote: >> Hmm. ?This and the image processing file from the current dev version >> of the package (2.1.3) are seemingly inconsistent? >> http://bioconductor.org/packages/2.8/bioc/vignettes/beadarray/inst/ doc/ImageProcessing.pdf >> >> Suggests that readIllumina always performs the backgroundNormalise process. >> >> http://bioconductor.org/packages/2.8/bioc/vignettes/beadarray/inst/ doc/beadlevel.pdf >> >> and what's below suggests that this is not the case anymore. >> >> Could you clarify? ?My goal in asking is to take data that's been read with: >> >> and use the neqc normalization process, which requires background >> correction not have been applied. >> >> Thanks, >> Anand >> >> On Tue, Dec 14, 2010 at 5:00 AM, ?<bioconductor-request at="" r-project.org=""> wrote: >>> Message: 3 >>> Date: Mon, 13 Dec 2010 14:24:45 +0000 >>> From: Mark Dunning <mark.dunning at="" gmail.com=""> >>> To: Changhe Fu <fuchanghe at="" gmail.com=""> >>> Cc: bioconductor at r-project.org >>> Subject: Re: [BioC] About package beadarray >>> Message-ID: >>> ? ? ? ?<aanlktik5pwuwyhtbwwbqsojcocw9f68ed5uk8feyfv=8 at="" mail.gmail.com=""> >>> Content-Type: text/plain; charset=ISO-8859-1 >>> >>> Hi, >>> >>> That function used to be in beadarray to implement the background >>> normalisation that occurs in Illumina's BeadStudio/GenomeStudio. >>> However, it has been shown in various studies to be inferior to other >>> methods. See the neqc (limma) or vst (lumi) alternatives available >>> through the normaliseIllumina function in beadarray. >>> >>> Mark >>> >>> >>> On Fri, Dec 10, 2010 at 11:07 PM, Changhe Fu <fuchanghe at="" gmail.com=""> wrote: >>>> Hi, All >>>> >>>> In previous version of package beadarray, I can use the function >>>> backgroundNormalise(). But in new one, I cann't find it. Is it included in >>>> some other fuction of normalising? >>>> >>>> Thank you >>>> >>>> -- >>>> Best, >>>> Changhe Fu >>>> >>>> ? ? ? ?[[alternative HTML version deleted]] >>>> >>>> _______________________________________________ >>>> Bioconductor mailing list >>>> Bioconductor at r-project.org >>>> https://stat.ethz.ch/mailman/listinfo/bioconductor >>>> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor >>>> >> >

Just adding to Mark's comments: if your data is bead-level data, it is still possible to perform a normexp background correction aided by negative controls. The nec() function in limma performs such a background correction. nec() is the same as neqc() except that it does not include the quantile between-array normalization and log2 transformation steps. Although neqc() and nec() are designed for pre-processing probe-level data (summarized data), I think using these functions to normalize bead-level data should give the result as good as applying them for the probe-level data, though no such comparisons have been performed. But if you want to work on the bead-level data and also want to use the normexp by control method to normalize them, this might be the way to go. Cheers, Wei On Jan 10, 2011, at 10:34 PM, Mark Dunning wrote: > Hi Anand, > > I think I understand the source of your confusion. There are two types > of correction applied to BeadArray data; one at the bead-level and one > at the summarized level. > > The bead-level text files created by Illumina's scanning software > report the intensity of each bead on the array. These have been > adjusted by subtracting a local background measure calculated using > nearby pixels on the image. For convenience, the readIllumina function > in beadarray will read these values, as these are the same values that > are analysed by Illumina's BeadStudio/GenomeStudio software. We find > them adequate for most purposes, although a more thorough analysis is > possible as described in the imageProcessing vignette. If you have > data exported from BeadStudio they will almost certainly have been > adjusted in this way. > > However, the backgroundNormalise method (that is now removed) referred > to in previous emails occurs on a per bead-type basis once data have > been summarized. It used to be the default normalization method in > BeadStudio. The quantities subtracted are negative controls > (bead-types with no target in the genome). As I mention in previous > posts there are viable alternatives to this method. The neqc method > you are interested in requires summarized data and (I suspect) > developed with GenomeStudio data in mind. Therefore it is Ok to use > data that have been read using readIllumina with this method. > > Hope this helps. Regards, > > Mark > > On Thu, Jan 6, 2011 at 8:37 PM, Anand Patel <acpatel at="" gmail.com=""> wrote: >> Hmm. This and the image processing file from the current dev version >> of the package (2.1.3) are seemingly inconsistent? >> http://bioconductor.org/packages/2.8/bioc/vignettes/beadarray/inst/ doc/ImageProcessing.pdf >> >> Suggests that readIllumina always performs the backgroundNormalise process. >> >> http://bioconductor.org/packages/2.8/bioc/vignettes/beadarray/inst/ doc/beadlevel.pdf >> >> and what's below suggests that this is not the case anymore. >> >> Could you clarify? My goal in asking is to take data that's been read with: >> >> and use the neqc normalization process, which requires background >> correction not have been applied. >> >> Thanks, >> Anand >> >> On Tue, Dec 14, 2010 at 5:00 AM, <bioconductor-request at="" r-project.org=""> wrote: >>> Message: 3 >>> Date: Mon, 13 Dec 2010 14:24:45 +0000 >>> From: Mark Dunning <mark.dunning at="" gmail.com=""> >>> To: Changhe Fu <fuchanghe at="" gmail.com=""> >>> Cc: bioconductor at r-project.org >>> Subject: Re: [BioC] About package beadarray >>> Message-ID: >>> <aanlktik5pwuwyhtbwwbqsojcocw9f68ed5uk8feyfv=8 at="" mail.gmail.com=""> >>> Content-Type: text/plain; charset=ISO-8859-1 >>> >>> Hi, >>> >>> That function used to be in beadarray to implement the background >>> normalisation that occurs in Illumina's BeadStudio/GenomeStudio. >>> However, it has been shown in various studies to be inferior to other >>> methods. See the neqc (limma) or vst (lumi) alternatives available >>> through the normaliseIllumina function in beadarray. >>> >>> Mark >>> >>> >>> On Fri, Dec 10, 2010 at 11:07 PM, Changhe Fu <fuchanghe at="" gmail.com=""> wrote: >>>> Hi, All >>>> >>>> In previous version of package beadarray, I can use the function >>>> backgroundNormalise(). But in new one, I cann't find it. Is it included in >>>> some other fuction of normalising? >>>> >>>> Thank you >>>> >>>> -- >>>> Best, >>>> Changhe Fu >>>> >>>> ? ? ? ?[[alternative HTML version deleted]] >>>> >>>> _______________________________________________ >>>> Bioconductor mailing list >>>> Bioconductor at r-project.org >>>> https://stat.ethz.ch/mailman/listinfo/bioconductor >>>> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor >>>> >> > > _______________________________________________ > Bioconductor mailing list > Bioconductor at r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor ______________________________________________________________________ The information in this email is confidential and intend...{{dropped:6}}