lumi, Illumina Methylation 450k, and robust methylation calls
2
0
Entering edit mode
Pan Du ▴ 440
@pan-du-4636
Last seen 10.3 years ago
Hi Tim The algorithm for methylation status estimation has not been well tested yet. That's why I did not put the details on the vignette. Now I am working on other parts of 450K methylation analysis, like DMR detection and annotation. I will work on methylation status estimation later when I have time. Sorry about it. Pan Date: Wed, 8 Jun 2011 12:38:58 +0100 From: Tim Rayner <tfrayner@gmail.com> To: Bioconductor <bioconductor@stat.math.ethz.ch> Subject: [BioC] lumi, Illumina Methylation 450k, and robust methylation calls Message-ID: <banlkti=7tu-5wtroxhk0qraucnevhnqqrw@mail.gmail.com> Content-Type: text/plain; charset="ISO-8859-1" Hi, I have recently started to use the lumi package to analyse some Illumina Human Methylation 450k data, and I have run into some problems which seem to revolve around division by zero in the gammaFitEM() function. I have adjusted the colour balance and quantile normalised as suggested in the vignette, but when I call gammaFitEM() the function complains (see the end of this email for a session dump). I've traced the likely cause of the error to zero values returned by these calls in gammaFitEM(): f1 <- dgamma(x - s[1], shape = k[1], scale = theta[1]) f2 <- dgamma(s[2] - x, shape = k[2], scale = theta[2]) The problem is that the z1 variable subsequently contains divisions by these zero values: z1 <- p[1] * f1/(p[1] * f1 + p[2] * f2) When z1 is later used in calls to sum() in many places, this obviously returns NaN which causes the function to raise an exception. I think I've got around this by editing the function and putting na.rm=TRUE in each of the relevant calls to sum(), and the generated plots look quite believable, but I can't be sure if that's actually a valid approach. Is there a better way to address this problem? Many thanks, Tim -- Tim Rayner Bioinformatician, Smith Lab, CIMR, University of Cambridge, U.K. ## session dump follows: > library(lumi) Loading required package: Biobase Welcome to Bioconductor Vignettes contain introductory material. To view, type 'browseVignettes()'. To cite Bioconductor, see 'citation("Biobase")' and for packages 'citation("pkgname")'. Loading required package: nleqslv KernSmooth 2.23 loaded Copyright M. P. Wand 1997-2009 Attaching package: 'lumi' > # data.c.quantile is the normalised MethyLumiM object. > fit<-gammaFitEM(exprs(data.c. quantile)[,1], plotMode=TRUE, verbose=TRUE) Initial estimation: k: 28 8 s: -10.17401 5.376681 theta: 0.2424133 0.4134306 p: 0.4264381 0.5735619 logLikelihood: -1135672 Error in if (abs(p.new[1] - p[1]) < tol) break : missing value where TRUE/FALSE needed > sessionInfo() R version 2.13.0 (2011-04-13) Platform: x86_64-apple-darwin9.8.0/x86_64 (64-bit) locale: [1] en_GB.UTF-8/en_GB.UTF-8/C/C/en_GB.UTF-8/en_GB.UTF-8 attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] lumi_2.4.0 nleqslv_1.8.5 Biobase_2.12.1 loaded via a namespace (and not attached): [1] affy_1.30.0 affyio_1.20.0 annotate_1.30.0 [4] AnnotationDbi_1.14.1 DBI_0.2-5 grid_2.13.0 [7] hdrcde_2.15 KernSmooth_2.23-5 lattice_0.19-26 [10] MASS_7.3-13 Matrix_0.999375-50 methylumi_1.8.0 [13] mgcv_1.7-6 nlme_3.1-101 preprocessCore_1.14.0 [16] RSQLite_0.9-4 tools_2.13.0 xtable_1.5-6 [[alternative HTML version deleted]]
lumi lumi • 1.6k views
ADD COMMENT •
0
Entering edit mode
@tiffany-morris-4695
Last seen 10.3 years ago
Tim Rayner <tfrayner at="" ...=""> writes: > I have recently started to use the lumi package to analyse some > Illumina Human Methylation 450k data, and I have run into some > problems which seem to revolve around division by zero in the > gammaFitEM() function. > I've got around this by editing the function and putting na.rm=TRUE in > each of the relevant calls to sum(), and the generated plots look > quite believable, but I can't be sure if that's actually a valid > approach. Is there a better way to address this problem? Hi, I am also having difficulty getting the gammaFitEM() function to work. Does anyone have a solution? Regards, Tiffany
ADD COMMENT •
0
Entering edit mode
Pan Du ▴ 440
@pan-du-4636
Last seen 10.3 years ago
Hi Ina and Tim For the methylation call part, I haven't work on that for a while because there are other priorities. For the color bias adjustment of 450K array, you need to use the development version, which perform color bias adjustment for both type I and II designs. Basically, it estimates the color bias based on type I design (methylated and unmenthylated measurements of the same CpG-site has the same color), which is the same as Infinium 27k, and then use the fitted curve to adjust probes with type II design (methylated and unmenthylated measurements of the same CpG-site has different colors). Pan Date: Mon, 13 Jun 2011 18:05:16 -0400 From: Ina Hoeschele <inah@vbi.vt.edu> To: Tim Rayner <tfrayner@gmail.com> Cc: Bioconductor <bioconductor@stat.math.ethz.ch> Subject: Re: [BioC] lumi, Illumina Methylation 450k, and robust methylation calls Message-ID: <9267536a-70f6-4582-a38f- 8d2288c2afed@zimbra> Content-Type: text/plain; charset="utf-8" << I have recently started to use the lumi package to analyse some Illumina Human Methylation 450k data, and I have run into some problems which seem to revolve around division by zero in the gammaFitEM() function. I have adjusted the colour balance and quantile normalised as suggested in the vignette >> Hi Pan and Tim (et al), this is in regard to an earlier email from Tim - is the colour balance adjustment performed for ALL probes or only for probes with Infinium I assay? Is the quantile normalization done for both Inf I and II together (rather than separately) in the current (development) version of lumi? I would be reluctant to do it that way, and I apologize if this is described somewhere and I misssed this informaiton. Thanks, Ina [[alternative HTML version deleted]]
ADD COMMENT •
0
Entering edit mode
thank you very much, Pan, so the color adjustment makes perfect sense to me now. Now for the background correction, lumi subtracts the median intensity of the negative control probes, while GenomeStudio subtracts the 5th percentile - does anyone know why? Thanks again, Ina ----- Original Message ----- From: "Pan Du" <dupan.mail@gmail.com> To: "Ina Hoeschele" <inah at="" vbi.vt.edu="">, "Tim Rayner" <tfrayner at="" gmail.com="">, bioconductor at r-project.org Sent: Tuesday, June 14, 2011 11:34:27 AM Subject: Re: [BioC] lumi, Illumina Methylation 450k, and robust methylation calls Hi Ina and Tim For the methylation call part, I haven't work on that for a while because there are other priorities. For the color bias adjustment of 450K array, you need to use the development version, which perform color bias adjustment for both type I and II designs. Basically, it estimates the color bias based on type I design (methylated and unmenthylated measurements of the same CpG-site has the same color), which is the same as Infinium 27k, and then use the fitted curve to adjust probes with type II design (methylated and unmenthylated measurements of the same CpG-site has different colors). Pan Date: Mon, 13 Jun 2011 18:05:16 -0400 From: Ina Hoeschele <inah@vbi.vt.edu> To: Tim Rayner <tfrayner at="" gmail.com=""> Cc: Bioconductor <bioconductor at="" stat.math.ethz.ch=""> Subject: Re: [BioC] lumi, Illumina Methylation 450k, and robust methylation calls Message-ID: <9267536a-70f6-4582-a38f- 8d2288c2afed at zimbra> Content-Type: text/plain; charset="utf-8" << I have recently started to use the lumi package to analyse some Illumina Human Methylation 450k data, and I have run into some problems which seem to revolve around division by zero in the gammaFitEM() function. I have adjusted the colour balance and quantile normalised as suggested in the vignette >> Hi Pan and Tim (et al), this is in regard to an earlier email from Tim - is the colour balance adjustment performed for ALL probes or only for probes with Infinium I assay? Is the quantile normalization done for both Inf I and II together (rather than separately) in the current (development) version of lumi? I would be reluctant to do it that way, and I apologize if this is described somewhere and I misssed this informaiton. Thanks, Ina
ADD REPLY •
0
Entering edit mode
Hi all, I'm running the methlumiB function in the lumi package on a Methylumi object of 8 450K meth bead chips - all it should be doing is to subtract the median intensity of my negative control data, but I am getting the error message below > batch2Meth.colQ.bg <- lumiMethyB(batch2Meth.colQ,method="bgAdjust2C",separateColor=F) Perform bgAdjust2C background correction ... Error: cannot allocate vector of size 355.6 Mb Does anyone have a hint for me (this is not hardware limitation!)? Thanks, Ina > sessionInfo() R version 2.14.0 Under development (unstable) (2011-05-19 r55967) Platform: x86_64-unknown-linux-gnu (64-bit) locale: [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8 [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 [7] LC_PAPER=C LC_NAME=C [9] LC_ADDRESS=C LC_TELEPHONE=C [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] affy_1.31.1 lattice_0.19-26 [3] IlluminaHumanMethylation450k.db_1.4.6 org.Hs.eg.db_2.5.0 [5] RSQLite_0.9-4 DBI_0.2-5 [7] AnnotationDbi_1.15.3 lumi_2.5.1 [9] nleqslv_1.8.5 methylumi_1.9.0 [11] Biobase_2.13.2 loaded via a namespace (and not attached): [1] affyio_1.21.1 annotate_1.31.0 grid_2.14.0 [4] hdrcde_2.15 KernSmooth_2.23-5 MASS_7.3-13 [7] Matrix_0.999375-50 mgcv_1.7-6 nlme_3.1-101 [10] preprocessCore_1.15.0 tools_2.14.0 xtable_1.5-6 ----- Original Message ----- From: "Ina Hoeschele" <inah@vbi.vt.edu> To: "Pan Du" <dupan.mail at="" gmail.com=""> Cc: bioconductor at r-project.org Sent: Wednesday, June 15, 2011 4:40:41 PM Subject: Re: [BioC] lumi, Illumina Methylation 450k, and robust methylation calls thank you very much, Pan, so the color adjustment makes perfect sense to me now. Now for the background correction, lumi subtracts the median intensity of the negative control probes, while GenomeStudio subtracts the 5th percentile - does anyone know why? Thanks again, Ina ----- Original Message ----- From: "Pan Du" <dupan.mail@gmail.com> To: "Ina Hoeschele" <inah at="" vbi.vt.edu="">, "Tim Rayner" <tfrayner at="" gmail.com="">, bioconductor at r-project.org Sent: Tuesday, June 14, 2011 11:34:27 AM Subject: Re: [BioC] lumi, Illumina Methylation 450k, and robust methylation calls Hi Ina and Tim For the methylation call part, I haven't work on that for a while because there are other priorities. For the color bias adjustment of 450K array, you need to use the development version, which perform color bias adjustment for both type I and II designs. Basically, it estimates the color bias based on type I design (methylated and unmenthylated measurements of the same CpG-site has the same color), which is the same as Infinium 27k, and then use the fitted curve to adjust probes with type II design (methylated and unmenthylated measurements of the same CpG-site has different colors). Pan Date: Mon, 13 Jun 2011 18:05:16 -0400 From: Ina Hoeschele <inah@vbi.vt.edu> To: Tim Rayner <tfrayner at="" gmail.com=""> Cc: Bioconductor <bioconductor at="" stat.math.ethz.ch=""> Subject: Re: [BioC] lumi, Illumina Methylation 450k, and robust methylation calls Message-ID: <9267536a-70f6-4582-a38f- 8d2288c2afed at zimbra> Content-Type: text/plain; charset="utf-8" << I have recently started to use the lumi package to analyse some Illumina Human Methylation 450k data, and I have run into some problems which seem to revolve around division by zero in the gammaFitEM() function. I have adjusted the colour balance and quantile normalised as suggested in the vignette >> Hi Pan and Tim (et al), this is in regard to an earlier email from Tim - is the colour balance adjustment performed for ALL probes or only for probes with Infinium I assay? Is the quantile normalization done for both Inf I and II together (rather than separately) in the current (development) version of lumi? I would be reluctant to do it that way, and I apologize if this is described somewhere and I misssed this informaiton. Thanks, Ina _______________________________________________ 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
ADD REPLY •
0
Entering edit mode
On Thu, Jun 16, 2011 at 2:29 PM, Ina Hoeschele <inah at="" vbi.vt.edu=""> wrote: > Hi all, > ?I'm running the methlumiB function in the lumi package on a Methylumi object of 8 450K meth bead chips - all it should be doing is to subtract the median intensity of my negative control data, but I am getting the error message below >> batch2Meth.colQ.bg <- lumiMethyB(batch2Meth.colQ,method="bgAdjust2C",separateColor=F) > Perform bgAdjust2C background correction ... > Error: cannot allocate vector of size 355.6 Mb Yes. Looks like you are out-of-memory. Make sure that you are using a clean workspace (ls()), so remove anything that you think you do not need. You are showing a "replacement" operation, but R doesn't do that in-place; instead it makes a copy and then replaces the original. Just out of curiosity, how much memory does the machine have? Sean > Does anyone have a hint for me (this is not hardware limitation!)? > > Thanks, Ina > >> sessionInfo() > R version 2.14.0 Under development (unstable) (2011-05-19 r55967) > Platform: x86_64-unknown-linux-gnu (64-bit) > > locale: > ?[1] LC_CTYPE=en_US.UTF-8 ? ? ? LC_NUMERIC=C > ?[3] LC_TIME=en_US.UTF-8 ? ? ? ?LC_COLLATE=en_US.UTF-8 > ?[5] LC_MONETARY=en_US.UTF-8 ? ?LC_MESSAGES=en_US.UTF-8 > ?[7] LC_PAPER=C ? ? ? ? ? ? ? ? LC_NAME=C > ?[9] LC_ADDRESS=C ? ? ? ? ? ? ? LC_TELEPHONE=C > [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C > > attached base packages: > [1] stats ? ? graphics ?grDevices utils ? ? datasets ?methods ? base > > other attached packages: > ?[1] affy_1.31.1 ? ? ? ? ? ? ? ? ? ? ? ? ? lattice_0.19-26 > ?[3] IlluminaHumanMethylation450k.db_1.4.6 org.Hs.eg.db_2.5.0 > ?[5] RSQLite_0.9-4 ? ? ? ? ? ? ? ? ? ? ? ? DBI_0.2-5 > ?[7] AnnotationDbi_1.15.3 ? ? ? ? ? ? ? ? ?lumi_2.5.1 > ?[9] nleqslv_1.8.5 ? ? ? ? ? ? ? ? ? ? ? ? methylumi_1.9.0 > [11] Biobase_2.13.2 > > loaded via a namespace (and not attached): > ?[1] affyio_1.21.1 ? ? ? ? annotate_1.31.0 ? ? ? grid_2.14.0 > ?[4] hdrcde_2.15 ? ? ? ? ? KernSmooth_2.23-5 ? ? MASS_7.3-13 > ?[7] Matrix_0.999375-50 ? ?mgcv_1.7-6 ? ? ? ? ? ?nlme_3.1-101 > [10] preprocessCore_1.15.0 tools_2.14.0 ? ? ? ? ?xtable_1.5-6 > > > ----- Original Message ----- > From: "Ina Hoeschele" <inah at="" vbi.vt.edu=""> > To: "Pan Du" <dupan.mail at="" gmail.com=""> > Cc: bioconductor at r-project.org > Sent: Wednesday, June 15, 2011 4:40:41 PM > Subject: Re: [BioC] lumi, Illumina Methylation 450k, ? ?and robust methylation calls > > thank you very much, Pan, so the color adjustment makes perfect sense to me now. Now for the background correction, lumi subtracts the median intensity of the negative control probes, while GenomeStudio subtracts the 5th percentile - does anyone know why? > Thanks again, Ina > > ----- Original Message ----- > From: "Pan Du" <dupan.mail at="" gmail.com=""> > To: "Ina Hoeschele" <inah at="" vbi.vt.edu="">, "Tim Rayner" <tfrayner at="" gmail.com="">, bioconductor at r-project.org > Sent: Tuesday, June 14, 2011 11:34:27 AM > Subject: Re: [BioC] lumi, Illumina Methylation 450k, and robust methylation calls > > Hi Ina and Tim > > For the methylation call part, I haven't work on that for a while because > there are other priorities. > > For the color bias adjustment of 450K array, you need to use the development > version, which perform color bias adjustment for both type I and II designs. > Basically, it estimates the color bias based on type I design (methylated > and unmenthylated measurements of the same CpG-site has ?the same color), > which is the same as Infinium 27k, and then use the fitted curve to adjust > probes with type II design (methylated and unmenthylated measurements of the > same CpG-site has different colors). > > > Pan > > > Date: Mon, 13 Jun 2011 18:05:16 -0400 > From: Ina Hoeschele <inah at="" vbi.vt.edu=""> > To: Tim Rayner <tfrayner at="" gmail.com=""> > Cc: Bioconductor <bioconductor at="" stat.math.ethz.ch=""> > Subject: Re: [BioC] lumi, Illumina Methylation 450k, ? ?and robust > ? ? ? methylation calls > Message-ID: <9267536a-70f6-4582-a38f- > 8d2288c2afed at zimbra> > Content-Type: text/plain; charset="utf-8" > > << > I have recently started to use the lumi package to analyse some > Illumina Human Methylation 450k data, and I have run into some > problems which seem to revolve around division by zero in the > gammaFitEM() function. I have adjusted the colour balance and quantile > normalised as suggested in the vignette >>> > > Hi Pan and Tim (et al), > ?this is in regard to an earlier email from Tim - is the colour balance > adjustment performed for ALL probes or only for probes with Infinium I > assay? Is the quantile normalization done for both Inf I and II together > (rather than separately) in the current (development) version of lumi? I > would be reluctant to do it that way, and I apologize if this is described > somewhere and I misssed this informaiton. > Thanks, Ina > > _______________________________________________ > 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 >
ADD REPLY •

Login before adding your answer.

Similar Posts
Loading Similar Posts
Traffic: 859 users visited in the last hour
Content Search
Users
Tags
Badges
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6