Dear All, Pardon my ignorance, but I am getting slightly desperate to coerce my methyLumi object to RGChannelSet. I tried all combinations of require(devtools) datrcg<-as(sampsM,'RGChannelSet') Error in methylumiToMinfi(from) : Cannot construct an RGChannelSet without full (OOB) intensities Any pointer would be very much appreciated. (I wonder if my inability to use the 'devel' is to blame...) Kind regards from the Netherlands Marco Marco PM Boks, Brain Center Rudolf Magnus, University Medical Centre Utrecht, HP. A.01.489, PO Box 85500, 3508 GA Utrecht, The Netherlands, Phone: +31 88 7556370 Fax: +31 88 7555509 E-mail: mboks@umcutrecht.nl <c.schubart@umcutrecht.nl> sessionInfo() R version 3.0.2 (2013-09-25) Platform: x86_64-unknown-linux-gnu (64-bit) locale: [1] LC_CTYPE=en_US.iso885915 LC_NUMERIC=C [3] LC_TIME=en_US.iso885915 LC_COLLATE=en_US.iso885915 [5] LC_MONETARY=en_US.iso885915 LC_MESSAGES=en_US.iso885915 [7] LC_PAPER=en_US.iso885915 LC_NAME=C [9] LC_ADDRESS=C LC_TELEPHONE=C [11] LC_MEASUREMENT=en_US.iso885915 LC_IDENTIFICATION=C attached base packages: [1] parallel stats graphics grDevices utils datasets methods [8] base other attached packages: [1] minfi_1.8.9 bumphunter_1.2.0 locfit_1.5-9.1 [4] iterators_1.0.6 foreach_1.4.1 Biostrings_2.30.1 [7] GenomicRanges_1.14.4 XVector_0.2.0 IRanges_1.20.6 [10] reshape_0.8.4 plyr_1.8 lattice_0.20-24 [13] methylumi_2.8.0 matrixStats_0.8.14 ggplot2_0.9.3.1 [16] reshape2_1.2.2 scales_0.2.3 Biobase_2.22.0 [19] BiocGenerics_0.8.0 devtools_1.4.1 BiocInstaller_1.12.0 loaded via a namespace (and not attached): [1] annotate_1.40.0 AnnotationDbi_1.24.0 base64_1.1 [4] beanplot_1.1 codetools_0.2-8 colorspace_1.2-4 [7] DBI_0.2-7 dichromat_2.0-0 digest_0.6.4 [10] doRNG_1.5.5 evaluate_0.5.1 genefilter_1.44.0 [13] grid_3.0.2 gtable_0.1.2 httr_0.2 [16] illuminaio_0.4.0 itertools_0.1-1 labeling_0.2 [19] limma_3.18.10 MASS_7.3-29 mclust_4.2 [22] memoise_0.1 multtest_2.18.0 munsell_0.4.2 [25] nlme_3.1-113 nor1mix_1.1-4 pkgmaker_0.17.4 [28] preprocessCore_1.24.0 proto_0.3-10 RColorBrewer_1.0-5 [31] RCurl_1.95-4.1 registry_0.2 R.methodsS3_1.6.1 [34] rngtools_1.2.3 RSQLite_0.11.4 siggenes_1.36.0 [37] splines_3.0.2 stats4_3.0.2 stringr_0.6.2 [40] survival_2.37-7 tools_3.0.2 whisker_0.3-2 [43] XML_3.98-1.1 xtable_1.7-1 2014-02-13 18:23 GMT+01:00 Tim Triche, Jr. <tim.triche@gmail.com>: > (cc:'ing the epigenomicsforum list due to a similar request there) > > You can now coerce (at least in devel) without the OOB probes (I checked > in an additional fix Monday morning, which was also needed to correct a > build issue), but you will still run into issues with minfi's approach. > There are at least two possibilities to get around this and still get > estimated cell counts. Note that I'd choose #2 (or a reasonable semblance > of it, i.e. quantile normalizing your data together with Reinius & Kere's > sorted leukocyte fractions) whenever feasible, but sometimes it isn't > feasible. Here's one way to deal with a lack of raw binary files (#1), and > some notes about Jaffe's approach (#2). > > 1) Use the modified version of the modified version of Houseman's code > (yes, that's in there twice!) > > require(devtools) > install_github('chromophobe','ttriche') > > ?composition ## I don't write man pages just for fun... > require(FlowSorted.Blood.450k) > counts <- getBloodCellCounts(your.GenomicRatioSet.coerced.from.lumi) > plotCellCounts(counts) > > > 2) Start from the IDATs, if you have them, in which case you can just run > the whole thing in minfi. Sometimes this is impossible (original authors > blithely ignore you, otherwise-extremely-helpful program officer for their > grant gets buried by a snowstorm, etc.) so I yanked out the essential bits > and stuffed them into a convenience function above. One of these days I'll > tidy up chromophobe enough to push it into BioC. But, if you have raw data > (and technical or biological replicates), it would be worth comparing the > two approaches to see whether your additional processing steps are helping > enough to keep them around. I may end up doing so (and perhaps comparing > both with EWASher). > > > Hope this helps, > > --t > > > On Sat, Feb 8, 2014 at 10:15 AM, Aileen Bahl <aileen.bahl@helsinki.fi>wrote: > >> Dear all, >> >> I used lumi to analyze my 450k data. Now, I want to use the Houseman >> algorithm to correct for the cell composition. As far as I read, the >> implementation in minfi covers several changes of the algorithm for the >> 450k platform. Therefore, I would like to use this package instead of >> changing the original code by hand. However, while trying to use minfi, I >> faced the problem of transforming my MethylumiM object to the required >> RGChannel object. I already found the coercions.R from Bioconductor but >> this gives me an error because the full OOB intensities are missing. Could >> you please provide some help on how to make this transformation in both >> directions (MethylumiM to RGchannel and the other way around)? >> >> Thanks in advance, >> Aileen >> >> _______________________________________________ >> 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]]

Thanks a lot! (It has finally landed!) I plan to coerce my methyLumiM object to RGChannel, normalise together with the sorted cell data and use getBloodCellCounts. Cheers Marco 2014-02-22 19:33 GMT+01:00 Tim Triche, Jr. <tim.triche@gmail.com>: > Yeah it's the devel version you want. I'll see about back porting > everything to 2.13 but I can't promise it will be super fast... You might > be better off installing the devel package from the BioC site (am I going > to get stoned to death for saying this?). > > Please note that you're still going to be hosed if you try to use > preprocessQuantile or estimateCellCounts on the coerced object. That's not > something I can fix; it's intrinsic to how minfi does these steps. You may > have to choose which steps are most important to you, since I don't think > lumi was ever modified to retain opposite-channel intensities of infinium I > probes > > Ymmv, > > --t > > On Feb 22, 2014, at 10:11 AM, Marco Boks <marcoboks@gmail.com> wrote: > > Dear All, > > Pardon my ignorance, but I am getting slightly desperate to coerce my > methyLumi object to RGChannelSet. > > I tried all combinations of > require(devtools) > datrcg<-as(sampsM,'RGChannelSet') > > Error in methylumiToMinfi(from) : > Cannot construct an RGChannelSet without full (OOB) intensities > > Any pointer would be very much appreciated. (I wonder if my inability to > use the 'devel' is to blame...) > > Kind regards from the Netherlands > > Marco > > Marco PM Boks, > > Brain Center Rudolf Magnus, > > University Medical Centre Utrecht, HP. A.01.489, > > PO Box 85500, 3508 GA Utrecht, > > The Netherlands, > > Phone: +31 88 7556370 > > Fax: +31 88 7555509 > > E-mail: mboks@umcutrecht.nl <c.schubart@umcutrecht.nl> > > > > > sessionInfo() > R version 3.0.2 (2013-09-25) > Platform: x86_64-unknown-linux-gnu (64-bit) > > locale: > [1] LC_CTYPE=en_US.iso885915 LC_NUMERIC=C > [3] LC_TIME=en_US.iso885915 LC_COLLATE=en_US.iso885915 > [5] LC_MONETARY=en_US.iso885915 LC_MESSAGES=en_US.iso885915 > [7] LC_PAPER=en_US.iso885915 LC_NAME=C > [9] LC_ADDRESS=C LC_TELEPHONE=C > [11] LC_MEASUREMENT=en_US.iso885915 LC_IDENTIFICATION=C > > attached base packages: > [1] parallel stats graphics grDevices utils datasets methods > [8] base > > other attached packages: > [1] minfi_1.8.9 bumphunter_1.2.0 locfit_1.5-9.1 > [4] iterators_1.0.6 foreach_1.4.1 Biostrings_2.30.1 > [7] GenomicRanges_1.14.4 XVector_0.2.0 IRanges_1.20.6 > [10] reshape_0.8.4 plyr_1.8 lattice_0.20-24 > [13] methylumi_2.8.0 matrixStats_0.8.14 ggplot2_0.9.3.1 > [16] reshape2_1.2.2 scales_0.2.3 Biobase_2.22.0 > [19] BiocGenerics_0.8.0 devtools_1.4.1 BiocInstaller_1.12.0 > > loaded via a namespace (and not attached): > [1] annotate_1.40.0 AnnotationDbi_1.24.0 base64_1.1 > [4] beanplot_1.1 codetools_0.2-8 colorspace_1.2-4 > [7] DBI_0.2-7 dichromat_2.0-0 digest_0.6.4 > [10] doRNG_1.5.5 evaluate_0.5.1 genefilter_1.44.0 > [13] grid_3.0.2 gtable_0.1.2 httr_0.2 > [16] illuminaio_0.4.0 itertools_0.1-1 labeling_0.2 > [19] limma_3.18.10 MASS_7.3-29 mclust_4.2 > [22] memoise_0.1 multtest_2.18.0 munsell_0.4.2 > [25] nlme_3.1-113 nor1mix_1.1-4 pkgmaker_0.17.4 > [28] preprocessCore_1.24.0 proto_0.3-10 RColorBrewer_1.0-5 > [31] RCurl_1.95-4.1 registry_0.2 R.methodsS3_1.6.1 > [34] rngtools_1.2.3 RSQLite_0.11.4 siggenes_1.36.0 > [37] splines_3.0.2 stats4_3.0.2 stringr_0.6.2 > [40] survival_2.37-7 tools_3.0.2 whisker_0.3-2 > [43] XML_3.98-1.1 xtable_1.7-1 > > > > 2014-02-13 18:23 GMT+01:00 Tim Triche, Jr. <tim.triche@gmail.com>: > >> (cc:'ing the epigenomicsforum list due to a similar request there) >> >> You can now coerce (at least in devel) without the OOB probes (I checked >> in an additional fix Monday morning, which was also needed to correct a >> build issue), but you will still run into issues with minfi's approach. >> There are at least two possibilities to get around this and still get >> estimated cell counts. Note that I'd choose #2 (or a reasonable semblance >> of it, i.e. quantile normalizing your data together with Reinius & Kere's >> sorted leukocyte fractions) whenever feasible, but sometimes it isn't >> feasible. Here's one way to deal with a lack of raw binary files (#1), and >> some notes about Jaffe's approach (#2). >> >> 1) Use the modified version of the modified version of Houseman's code >> (yes, that's in there twice!) >> >> require(devtools) >> install_github('chromophobe','ttriche') >> >> ?composition ## I don't write man pages just for fun... >> require(FlowSorted.Blood.450k) >> counts <- getBloodCellCounts(your.GenomicRatioSet.coerced.from.lumi) >> plotCellCounts(counts) >> >> >> 2) Start from the IDATs, if you have them, in which case you can just run >> the whole thing in minfi. Sometimes this is impossible (original authors >> blithely ignore you, otherwise-extremely-helpful program officer for their >> grant gets buried by a snowstorm, etc.) so I yanked out the essential bits >> and stuffed them into a convenience function above. One of these days I'll >> tidy up chromophobe enough to push it into BioC. But, if you have raw data >> (and technical or biological replicates), it would be worth comparing the >> two approaches to see whether your additional processing steps are helping >> enough to keep them around. I may end up doing so (and perhaps comparing >> both with EWASher). >> >> >> Hope this helps, >> >> --t >> >> >> On Sat, Feb 8, 2014 at 10:15 AM, Aileen Bahl <aileen.bahl@helsinki.fi>wrote: >> >>> Dear all, >>> >>> I used lumi to analyze my 450k data. Now, I want to use the Houseman >>> algorithm to correct for the cell composition. As far as I read, the >>> implementation in minfi covers several changes of the algorithm for the >>> 450k platform. Therefore, I would like to use this package instead of >>> changing the original code by hand. However, while trying to use minfi, I >>> faced the problem of transforming my MethylumiM object to the required >>> RGChannel object. I already found the coercions.R from Bioconductor but >>> this gives me an error because the full OOB intensities are missing. Could >>> you please provide some help on how to make this transformation in both >>> directions (MethylumiM to RGchannel and the other way around)? >>> >>> Thanks in advance, >>> Aileen >>> >>> _______________________________________________ >>> 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]]

Installed the devel, and tried; What am I doing wrong? Thank you for your patience.. Marco >datrcg<-as(sampsM,'RGChannelSet') Error in methylumiToMinfi(from) : Cannot construct an RGChannelSet without full (OOB) intensities > sessionInfo() R Under development (unstable) (2014-02-22 r65060) Platform: i386-w64-mingw32/i386 (32-bit) locale: [1] LC_COLLATE=English_United States.1252 [2] LC_CTYPE=English_United States.1252 [3] LC_MONETARY=English_United States.1252 [4] LC_NUMERIC=C [5] LC_TIME=English_United States.1252 attached base packages: [1] parallel stats graphics grDevices utils datasets methods [8] base other attached packages: [1] lumi_2.15.4 devtools_1.4.1 methylumi_2.9.12 [4] minfi_1.9.11 bumphunter_1.3.7 locfit_1.5-9.1 [7] iterators_1.0.6 foreach_1.4.1 Biostrings_2.31.14 [10] XVector_0.3.7 GenomicRanges_1.15.31 IRanges_1.21.32 [13] lattice_0.20-24 matrixStats_0.8.14 ggplot2_0.9.3.1 [16] reshape2_1.2.2 scales_0.2.3 Biobase_2.23.5 [19] BiocGenerics_0.9.3 BiocInstaller_1.13.3 loaded via a namespace (and not attached): [1] affy_1.41.3 affyio_1.31.0 [3] annotate_1.41.1 AnnotationDbi_1.25.9 [5] base64_1.1 BatchJobs_1.2 [7] BBmisc_1.5 beanplot_1.1 [9] BiocParallel_0.5.8 biomaRt_2.19.3 [11] bitops_1.0-6 brew_1.0-6 [13] BSgenome_1.31.11 codetools_0.2-8 [15] colorspace_1.2-4 DBI_0.2-7 [17] dichromat_2.0-0 digest_0.6.4 [19] doRNG_1.5.5 evaluate_0.5.1 [21] fail_1.2 genefilter_1.45.1 [23] GenomicAlignments_0.99.24 GenomicFeatures_1.15.7 [25] grid_3.1.0 gtable_0.1.2 [27] httr_0.2 illuminaio_0.5.5 [29] KernSmooth_2.23-10 labeling_0.2 [31] limma_3.19.20 MASS_7.3-29 [33] Matrix_1.1-2 mclust_4.2 [35] memoise_0.1 mgcv_1.7-28 [37] multtest_2.19.1 munsell_0.4.2 [39] nleqslv_2.1.1 nlme_3.1-113 [41] nor1mix_1.1-4 pkgmaker_0.17.4 [43] plyr_1.8 preprocessCore_1.25.5 [45] proto_0.3-10 R.methodsS3_1.6.1 [47] RColorBrewer_1.0-5 RCurl_1.95-4.1 [49] registry_0.2 reshape_0.8.4 [51] rngtools_1.2.3 Rsamtools_1.15.28 [53] RSQLite_0.11.4 rtracklayer_1.23.12 [55] sendmailR_1.1-2 siggenes_1.37.1 [57] splines_3.1.0 stats4_3.1.0 [59] stringr_0.6.2 survival_2.37-7 [61] tools_3.1.0 whisker_0.3-2 [63] XML_3.98-1.1 xtable_1.7-1 [65] zlibbioc_1.9.0 2014-02-22 19:33 GMT+01:00 Tim Triche, Jr. <tim.triche@gmail.com>: > Yeah it's the devel version you want. I'll see about back porting > everything to 2.13 but I can't promise it will be super fast... You might > be better off installing the devel package from the BioC site (am I going > to get stoned to death for saying this?). > > Please note that you're still going to be hosed if you try to use > preprocessQuantile or estimateCellCounts on the coerced object. That's not > something I can fix; it's intrinsic to how minfi does these steps. You may > have to choose which steps are most important to you, since I don't think > lumi was ever modified to retain opposite-channel intensities of infinium I > probes > > Ymmv, > > --t > > On Feb 22, 2014, at 10:11 AM, Marco Boks <marcoboks@gmail.com> wrote: > > Dear All, > > Pardon my ignorance, but I am getting slightly desperate to coerce my > methyLumi object to RGChannelSet. > > I tried all combinations of > require(devtools) > datrcg<-as(sampsM,'RGChannelSet') > > Error in methylumiToMinfi(from) : > Cannot construct an RGChannelSet without full (OOB) intensities > > Any pointer would be very much appreciated. (I wonder if my inability to > use the 'devel' is to blame...) > > Kind regards from the Netherlands > > Marco > > Marco PM Boks, > > Brain Center Rudolf Magnus, > > University Medical Centre Utrecht, HP. A.01.489, > > PO Box 85500, 3508 GA Utrecht, > > The Netherlands, > > Phone: +31 88 7556370 > > Fax: +31 88 7555509 > > E-mail: mboks@umcutrecht.nl <c.schubart@umcutrecht.nl> > > > > > sessionInfo() > R version 3.0.2 (2013-09-25) > Platform: x86_64-unknown-linux-gnu (64-bit) > > locale: > [1] LC_CTYPE=en_US.iso885915 LC_NUMERIC=C > [3] LC_TIME=en_US.iso885915 LC_COLLATE=en_US.iso885915 > [5] LC_MONETARY=en_US.iso885915 LC_MESSAGES=en_US.iso885915 > [7] LC_PAPER=en_US.iso885915 LC_NAME=C > [9] LC_ADDRESS=C LC_TELEPHONE=C > [11] LC_MEASUREMENT=en_US.iso885915 LC_IDENTIFICATION=C > > attached base packages: > [1] parallel stats graphics grDevices utils datasets methods > [8] base > > other attached packages: > [1] minfi_1.8.9 bumphunter_1.2.0 locfit_1.5-9.1 > [4] iterators_1.0.6 foreach_1.4.1 Biostrings_2.30.1 > [7] GenomicRanges_1.14.4 XVector_0.2.0 IRanges_1.20.6 > [10] reshape_0.8.4 plyr_1.8 lattice_0.20-24 > [13] methylumi_2.8.0 matrixStats_0.8.14 ggplot2_0.9.3.1 > [16] reshape2_1.2.2 scales_0.2.3 Biobase_2.22.0 > [19] BiocGenerics_0.8.0 devtools_1.4.1 BiocInstaller_1.12.0 > > loaded via a namespace (and not attached): > [1] annotate_1.40.0 AnnotationDbi_1.24.0 base64_1.1 > [4] beanplot_1.1 codetools_0.2-8 colorspace_1.2-4 > [7] DBI_0.2-7 dichromat_2.0-0 digest_0.6.4 > [10] doRNG_1.5.5 evaluate_0.5.1 genefilter_1.44.0 > [13] grid_3.0.2 gtable_0.1.2 httr_0.2 > [16] illuminaio_0.4.0 itertools_0.1-1 labeling_0.2 > [19] limma_3.18.10 MASS_7.3-29 mclust_4.2 > [22] memoise_0.1 multtest_2.18.0 munsell_0.4.2 > [25] nlme_3.1-113 nor1mix_1.1-4 pkgmaker_0.17.4 > [28] preprocessCore_1.24.0 proto_0.3-10 RColorBrewer_1.0-5 > [31] RCurl_1.95-4.1 registry_0.2 R.methodsS3_1.6.1 > [34] rngtools_1.2.3 RSQLite_0.11.4 siggenes_1.36.0 > [37] splines_3.0.2 stats4_3.0.2 stringr_0.6.2 > [40] survival_2.37-7 tools_3.0.2 whisker_0.3-2 > [43] XML_3.98-1.1 xtable_1.7-1 > > > > 2014-02-13 18:23 GMT+01:00 Tim Triche, Jr. <tim.triche@gmail.com>: > >> (cc:'ing the epigenomicsforum list due to a similar request there) >> >> You can now coerce (at least in devel) without the OOB probes (I checked >> in an additional fix Monday morning, which was also needed to correct a >> build issue), but you will still run into issues with minfi's approach. >> There are at least two possibilities to get around this and still get >> estimated cell counts. Note that I'd choose #2 (or a reasonable semblance >> of it, i.e. quantile normalizing your data together with Reinius & Kere's >> sorted leukocyte fractions) whenever feasible, but sometimes it isn't >> feasible. Here's one way to deal with a lack of raw binary files (#1), and >> some notes about Jaffe's approach (#2). >> >> 1) Use the modified version of the modified version of Houseman's code >> (yes, that's in there twice!) >> >> require(devtools) >> install_github('chromophobe','ttriche') >> >> ?composition ## I don't write man pages just for fun... >> require(FlowSorted.Blood.450k) >> counts <- getBloodCellCounts(your.GenomicRatioSet.coerced.from.lumi) >> plotCellCounts(counts) >> >> >> 2) Start from the IDATs, if you have them, in which case you can just run >> the whole thing in minfi. Sometimes this is impossible (original authors >> blithely ignore you, otherwise-extremely-helpful program officer for their >> grant gets buried by a snowstorm, etc.) so I yanked out the essential bits >> and stuffed them into a convenience function above. One of these days I'll >> tidy up chromophobe enough to push it into BioC. But, if you have raw data >> (and technical or biological replicates), it would be worth comparing the >> two approaches to see whether your additional processing steps are helping >> enough to keep them around. I may end up doing so (and perhaps comparing >> both with EWASher). >> >> >> Hope this helps, >> >> --t >> >> >> On Sat, Feb 8, 2014 at 10:15 AM, Aileen Bahl <aileen.bahl@helsinki.fi>wrote: >> >>> Dear all, >>> >>> I used lumi to analyze my 450k data. Now, I want to use the Houseman >>> algorithm to correct for the cell composition. As far as I read, the >>> implementation in minfi covers several changes of the algorithm for the >>> 450k platform. Therefore, I would like to use this package instead of >>> changing the original code by hand. However, while trying to use minfi, I >>> faced the problem of transforming my MethylumiM object to the required >>> RGChannel object. I already found the coercions.R from Bioconductor but >>> this gives me an error because the full OOB intensities are missing. Could >>> you please provide some help on how to make this transformation in both >>> directions (MethylumiM to RGchannel and the other way around)? >>> >>> Thanks in advance, >>> Aileen >>> >>> _______________________________________________ >>> 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]]