Hi, I am relatively new to the analysis of methylation data (Illumina 450k), and I have a question on how the methylation values are calculated in minfi. It did not became clear to me even after reading the help pages. I am asking because for some probes I observe that both Meth and Unmeth values of the non-BG corrected and non-normalized data are exactly zero, whereas the Green and Red channels are not. As a result NaN's are returned for the beta values, which interferes with the clustering of the raw data (QC-ing). - Could this be due to background correction, even though this should not be performed (i.e. this behavior is unintentional)?? - How to avoid these zero's? Likely by using an offset, but how to best do this? Thanks for any clarification! Guido Example: > rgSet <- read.450k.exp(base = myDir, targets = targets) > rgSet RGChannelSet (storageMode: lockedEnvironment) assayData: 622399 features, 20 samples element names: Green, Red phenoData sampleNames: PBMC_4 PBMC_13 ... PBMC_24 (20 total) varLabels: Sample_Name Sample_Group ... filenames (8 total) varMetadata: labelDescription Annotation array: IlluminaHumanMethylation450k annotation: ilmn.v1.2 > # Lets check the probeset on row 371,399, since that is one of 80 returning NaN's # First extract Green and Red channel intensities for 1st 2 arrays: > getGreen(rgSet)[371399,1:2] PBMC_4 PBMC_13 295 325 > getRed(rgSet)[371399,1:2] PBMC_4 PBMC_13 115 253 > # Now convert into methylation signal without any BG correction and normalization > mset <- preprocessRaw(rgSet) > mset MethylSet (storageMode: lockedEnvironment) assayData: 485512 features, 20 samples element names: Meth, Unmeth phenoData sampleNames: PBMC_4 PBMC_13 ... PBMC_24 (20 total) varLabels: Sample_Name Sample_Group ... filenames (8 total) varMetadata: labelDescription Annotation array: IlluminaHumanMethylation450k annotation: ilmn.v1.2 Preprocessing Method: Raw (no normalization or bg correction) minfi version: 1.6.0 Manifest version: 0.4.0 > # Extract methylation values for indicated probeset for 1st two arrays: > getMeth(mset)[371399,1:2] PBMC_4 PBMC_13 0 2660 > getUnmeth(mset)[371399,1:2] PBMC_4 PBMC_13 0 181 > #^ Values=0 is returned, why is this? # convert / extract betas: > betas = getBeta(mset) > betas[371399,1:2] PBMC_4 PBMC_13 NaN 0.93629 > # NaN's are introduced... > sessionInfo() R version 3.0.0 Patched (2013-04-15 r62590) Platform: x86_64-w64-mingw32/x64 (64-bit) locale: [1] LC_COLLATE=English_United States.1252 LC_CTYPE=English_United States.1252 LC_MONETARY=English_United States.1252 LC_NUMERIC=C [5] LC_TIME=English_United States.1252 attached base packages: [1] parallel stats graphics grDevices utils datasets methods base other attached packages: [1] IlluminaHumanMethylation450kmanifest_0.4.0 limma_3.16.3 minfi_1.6.0 [4] Biostrings_2.28.0 GenomicRanges_1.12.2 IRanges_1.18.1 [7] reshape_0.8.4 plyr_1.8 lattice_0.20-15 [10] Biobase_2.20.0 BiocGenerics_0.6.0 loaded via a namespace (and not attached): [1] beanplot_1.1 grid_3.0.0 illuminaio_0.2.0 MASS_7.3-26 matrixStats_0.8.1 mclust_4.1 multtest_2.16.0 [8] nor1mix_1.1-4 preprocessCore_1.22.0 R.methodsS3_1.4.2 RColorBrewer_1.0-5 siggenes_1.34.0 splines_3.0.0 stats4_3.0.0 [15] survival_2.37-4 tools_3.0.0 > [[alternative HTML version deleted]]

You can use getBeta(object, type = "illumina") to use the offset recommended by Illumina. There is _not_ a 1-1 correspondence between rows in the MethylSet and rows in the RGChannelSet, as your code suggests. Specifically, the Red/Green intensities you print does not correspond to the rows of the MethylSet which has NAs. Kasper On Mon, May 27, 2013 at 3:25 PM, Hooiveld, Guido <guido.hooiveld@wur.nl>wrote: > Hi, > I am relatively new to the analysis of methylation data (Illumina 450k), > and I have a question on how the methylation values are calculated in > minfi. It did not became clear to me even after reading the help pages. > I am asking because for some probes I observe that both Meth and Unmeth > values of the non-BG corrected and non-normalized data are exactly zero, > whereas the Green and Red channels are not. As a result NaN's are returned > for the beta values, which interferes with the clustering of the raw data > (QC-ing). > - Could this be due to background correction, even though this should not > be performed (i.e. this behavior is unintentional)?? > - How to avoid these zero's? Likely by using an offset, but how to best do > this? > > Thanks for any clarification! > Guido > > > Example: > > rgSet <- read.450k.exp(base = myDir, targets = targets) > > rgSet > RGChannelSet (storageMode: lockedEnvironment) > assayData: 622399 features, 20 samples > element names: Green, Red > phenoData > sampleNames: PBMC_4 PBMC_13 ... PBMC_24 (20 total) > varLabels: Sample_Name Sample_Group ... filenames (8 total) > varMetadata: labelDescription > Annotation > array: IlluminaHumanMethylation450k > annotation: ilmn.v1.2 > > > # Lets check the probeset on row 371,399, since that is one of 80 > returning NaN's > # First extract Green and Red channel intensities for 1st 2 arrays: > > getGreen(rgSet)[371399,1:2] > PBMC_4 PBMC_13 > 295 325 > > getRed(rgSet)[371399,1:2] > PBMC_4 PBMC_13 > 115 253 > > > # Now convert into methylation signal without any BG correction and > normalization > > mset <- preprocessRaw(rgSet) > > mset > MethylSet (storageMode: lockedEnvironment) > assayData: 485512 features, 20 samples > element names: Meth, Unmeth > phenoData > sampleNames: PBMC_4 PBMC_13 ... PBMC_24 (20 total) > varLabels: Sample_Name Sample_Group ... filenames (8 total) > varMetadata: labelDescription > Annotation > array: IlluminaHumanMethylation450k > annotation: ilmn.v1.2 > Preprocessing > Method: Raw (no normalization or bg correction) > minfi version: 1.6.0 > Manifest version: 0.4.0 > > > # Extract methylation values for indicated probeset for 1st two arrays: > > getMeth(mset)[371399,1:2] > PBMC_4 PBMC_13 > 0 2660 > > getUnmeth(mset)[371399,1:2] > PBMC_4 PBMC_13 > 0 181 > > > #^ Values=0 is returned, why is this? > > # convert / extract betas: > > betas = getBeta(mset) > > betas[371399,1:2] > PBMC_4 PBMC_13 > NaN 0.93629 > > > # NaN's are introduced... > > > > sessionInfo() > R version 3.0.0 Patched (2013-04-15 r62590) > Platform: x86_64-w64-mingw32/x64 (64-bit) > > locale: > [1] LC_COLLATE=English_United States.1252 LC_CTYPE=English_United > States.1252 LC_MONETARY=English_United States.1252 LC_NUMERIC=C > [5] LC_TIME=English_United States.1252 > > attached base packages: > [1] parallel stats graphics grDevices utils datasets methods > base > > other attached packages: > [1] IlluminaHumanMethylation450kmanifest_0.4.0 limma_3.16.3 > minfi_1.6.0 > [4] Biostrings_2.28.0 GenomicRanges_1.12.2 > IRanges_1.18.1 > [7] reshape_0.8.4 plyr_1.8 > lattice_0.20-15 > [10] Biobase_2.20.0 BiocGenerics_0.6.0 > > loaded via a namespace (and not attached): > [1] beanplot_1.1 grid_3.0.0 illuminaio_0.2.0 > MASS_7.3-26 matrixStats_0.8.1 mclust_4.1 > multtest_2.16.0 > [8] nor1mix_1.1-4 preprocessCore_1.22.0 R.methodsS3_1.4.2 > RColorBrewer_1.0-5 siggenes_1.34.0 splines_3.0.0 > stats4_3.0.0 > [15] survival_2.37-4 tools_3.0.0 > > > > > [[alternative HTML version deleted]] > > _______________________________________________ > Bioconductor mailing list > Bioconductor@r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor > [[alternative HTML version deleted]]