Dear AlI, I am very appreciate for your kidly help. I am working on the Illumina methylation27 data, I would like to combine two data from two plates. According to the Illumina's recommendation, "Background intensity computed from a set of negative controls was subtracted from each analytical data point. The ratio of fluorescent signals was then computed from the two alleles â = (max(M, 0))/(|U| + |M| + 100)." But a portion of my samples don't have the negative_control_Grn_Arg_intensity, some of them even don't have the negative_control_red_Arg_intensity. In that case, how can I estimate the background intensity? I also know that there is a dye shifting which might effect the combination of two dataset. How do I evaluate this dye problem on one plate? How to correct them? Which software or tools should I consider? Thank you. Grace [[alternative HTML version deleted]]

Hi Grace The developing version of lumi package has implemented functions of background adjustment and color balance adjustment functions for Illumina Infinium methylation microarrays. Please check the vignette, methylationAnalysis.pdf for more details. Here is the link: http://bioconductor.org/packages/2.7/bioc/html/lumi.html Pan On 9/30/10 5:00 AM, "bioconductor-request at stat.math.ethz.ch" <bioconductor-request at="" stat.math.ethz.ch=""> wrote: > Message: 27 > Date: Thu, 30 Sep 2010 11:40:20 +0800 > From: zhiqun tang <sisitzq at="" gmail.com=""> > To: bioconductor at stat.math.ethz.ch > Subject: [BioC] Illumina Methylation data > Message-ID: > <aanlktik_fzr5+ftaac5z_qz4e4z-_xbte7eiy1z_r8nr at="" mail.gmail.com=""> > Content-Type: text/plain > > Dear AlI, > I am very appreciate for your kidly help. > I am working on the Illumina methylation27 data, I would like to combine two > data from two plates. According to the Illumina's recommendation, "Background > intensity computed from a set of negative controls was subtracted from each > analytical data point. The ratio of fluorescent signals was then computed > from the two alleles ? = (max(M, 0))/(|U| + |M| + 100)." > But a portion of my samples don't have the > negative_control_Grn_Arg_intensity, some of them even don't have the > negative_control_red_Arg_intensity. In that case, how can I estimate the > background intensity? > I also know that there is a dye shifting which might effect the combination > of two dataset. How do I evaluate this dye problem on one plate? How to > correct them? Which software or tools should I consider? Thank you. > > Grace > > [[alternative HTML version deleted]] >

Hi Zhiqun, do you have evidence of nonlinear dye bias effects? remember, the probes on the HM27k arrays are paired within the same channel, and all recommended summary statistics for the assay are based on ratios. For any pair of numbers (M, U), both the beta value and the M-value will be identical before and after rescaling the intensities. (So for example, a ratio of 1.07 for Cy3/Cy5 is typical... 1.07*M / (1.07*M + 1.07*U) gives you the same beta, and log2(1.07*M / 1.07*U ) gives you the same logratio, as not bothering with any rescaling at all.) I haven't seen evidence that methods such as quantile normalization are particularly effective at removing any residual nonlinearities. Sensible background correction is important for position-specific effect amelioration on these chips. I wrote some patches for methyLumi that can take advantage of the various control probes to handle dye bias and background correction; as far as I am concerned, if the lab will not provide you with (at least) the summarized probe-level intensities, including the control probes, then you should ask for your money back. The assay generates a lot of information via the control probes. But... What I'd worry a lot more about if I were you is the plate-specific hybridization effects. Regardless of whether you use log-ratios or beta-values for analysis, if you have replicates and/or randomization, fitting a variance components model will quickly show where the worst effects are. All of the above will get merged into the 'lumi' package going forward, which I assume will soon have a controlData slot for methylation data similar to the one for expression data. 'addControlData2lumiMethy' would be a good idea, maybe I need to write this as a patch? (Pan?) You might find it easier to work initially with methyLumi, and then transition to lumi in the near future. If you want a quick fix, you can use something like Teschendorff's iSVA to try and correct some of the effects; I expect to see a decent paper on normalization in the very near future, as well :-) since most of the assumptions for expression arrays are false for methylation arrays, and the most sensible ideas I have seen are all coming from a genotyping perspective (crlmm, CHARM, etc.) On Wed, Sep 29, 2010 at 8:40 PM, zhiqun tang <sisitzq@gmail.com> wrote: > Dear AlI, > I am very appreciate for your kidly help. > I am working on the Illumina methylation27 data, I would like to combine > two > data from two plates. According to the Illumina's recommendation, > "Background > intensity computed from a set of negative controls was subtracted from each > analytical data point. The ratio of fluorescent signals was then computed > from the two alleles â = (max(M, 0))/(|U| + |M| + 100)." > But a portion of my samples don't have the > negative_control_Grn_Arg_intensity, some of them even don't have the > negative_control_red_Arg_intensity. In that case, how can I estimate the > background intensity? > I also know that there is a dye shifting which might effect the combination > of two dataset. How do I evaluate this dye problem on one plate? How to > correct them? Which software or tools should I consider? Thank you. > > Grace > > [[alternative HTML version deleted]] > > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor > > -- With four parameters I can fit an elephant, and with five I can make him wiggle his trunk. John von Neumann [[alternative HTML version deleted]]