beadarray: mRNA QC help
1
0
Entering edit mode
@abhishek-pratap-6167
Last seen 9.6 years ago
Hi All I am trying to analyze some mRNA illumina bead level data through bead array. Based on the detection p-values(plot attached): if I read it right significant # probes have no expression which concerns me but I could have easily missed some important step. I am attaching the two box plots 1.) intensity and 2.) probe level detection p-values + code Let me know your opinion. Thanks! -Abhi beadlevel_data <- readIllumina(dir=dataDir, useImages=F, illuminaAnnotation="Humanv4") #intensity boxplot boxplot(beadlevel_data, transFun = logGreenChannelTransform, col = "green", ylab = expression(log[2](intensity)), las=2, outline=FALSE, main= "Array intensities") #summarize : detection p-val boxplot BSData <- summarize(beadlevel_data) det = calculateDetection(BSData) boxplot(det, main='dset1_mRNA_probe_detection_pvals (Null: there is no expression) ', ylab='p-value(detection score)',xaxt='n', xlab='individual samples/array' ) -------------- next part -------------- A non-text attachment was scrubbed... Name: dset1_mRNA_array_intensities.png Type: image/png Size: 269613 bytes Desc: not available URL: <https: stat.ethz.ch="" pipermail="" bioconductor="" attachments="" 20140123="" f4138bba="" attachment.png=""> -------------- next part -------------- A non-text attachment was scrubbed... Name: dset1_mRNA_probe_detection_pvals.png Type: image/png Size: 87786 bytes Desc: not available URL: <https: stat.ethz.ch="" pipermail="" bioconductor="" attachments="" 20140123="" f4138bba="" attachment-0001.png="">
probe probe • 1.2k views
ADD COMMENT
0
Entering edit mode
Levi Waldron ▴ 80
@levi-waldron-6357
Last seen 9.6 years ago
Dear Abhi, it looks to me like your chips had low hybridization, based both on the low percentage of probes detected and on the low interquartile range of intensities. Fig. 2 from PMID23136189 (admittedly my publication) shows the range of IQRs for samples within several published studies using Illumina BeadArrays for FFPE tissues. The normal IQR is around 2, which corresponds to around 50-60% present at p<0.05. Your typical IQR is around 1, and I can't directly read percent present but it looks around 20-30%. I would be concerned, and do some additional checks: 1) if you have technical replicates, check them by MA plot and sample-wise correlations, 2) MA plots against the median pseudochip, 3) check for detection and high intensity of the positive control probes (labelled CY3_HYB and HOUSEKEEPING on Illumina chips I've looked at), and 4) run the ffpe::sortedIqrPlot() function to see whether IQR or percent present are related to correlation to median pseudochip in your experiment. Are you using FFPE specimens? Sincerely, Levi On Thu, Jan 23, 2014 at 6:38 PM, Abhishek Pratap <apratap@sagebase.org>wrote: > Hi All > > I am trying to analyze some mRNA illumina bead level data through bead > array. Based on the detection p-values(plot attached): if I read it > right significant # probes > have no expression which concerns me but I could have easily missed > some important step. > > I am attaching the two box plots 1.) intensity and 2.) probe level > detection p-values + code > > Let me know your opinion. > > Thanks! > -Abhi > > beadlevel_data <- readIllumina(dir=dataDir, useImages=F, > illuminaAnnotation="Humanv4") > > #intensity boxplot > boxplot(beadlevel_data, transFun = logGreenChannelTransform, col = "green", > ylab = expression(log[2](intensity)), las=2, outline=FALSE, > main= "Array intensities") > > #summarize : detection p-val boxplot > BSData <- summarize(beadlevel_data) > det = calculateDetection(BSData) > boxplot(det, main='dset1_mRNA_probe_detection_pvals (Null: there is no > expression) ', ylab='p-value(detection score)',xaxt='n', > xlab='individual samples/array' ) > > _______________________________________________ > 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 > -- Levi Waldron Assistant Professor of Biostatistics City University of New York School of Public Health, Hunter College 2180 3rd Ave Rm 538 New York NY 10035-4003 phone: 212-396-7747 [[alternative HTML version deleted]]
ADD COMMENT
0
Entering edit mode
Hi Levi Thanks for the quick feedback. I am still following up on few of the things that you recommended. I checked with the collab and these were not FFPE specimens. Also I did a quick signal to noise ratio calculation (95% / 5% intensity quantiles ). I am attaching the plot. May be that indicates something for the Illumina bead array platform that I am not aware of. To me it looks skewed. Cheers! -Abhi On Thu, Jan 23, 2014 at 4:26 PM, Levi Waldron <levi.waldron at="" hunter.cuny.edu=""> wrote: > Dear Abhi, > > it looks to me like your chips had low hybridization, based both on the low > percentage of probes detected and on the low interquartile range of > intensities. Fig. 2 from PMID23136189 (admittedly my publication) shows the > range of IQRs for samples within several published studies using Illumina > BeadArrays for FFPE tissues. The normal IQR is around 2, which corresponds > to around 50-60% present at p<0.05. Your typical IQR is around 1, and I > can't directly read percent present but it looks around 20-30%. I would be > concerned, and do some additional checks: 1) if you have technical > replicates, check them by MA plot and sample-wise correlations, 2) MA plots > against the median pseudochip, 3) check for detection and high intensity of > the positive control probes (labelled CY3_HYB and HOUSEKEEPING on Illumina > chips I've looked at), and 4) run the ffpe::sortedIqrPlot() function to see > whether IQR or percent present are related to correlation to median > pseudochip in your experiment. Are you using FFPE specimens? > > Sincerely, > Levi > > > > On Thu, Jan 23, 2014 at 6:38 PM, Abhishek Pratap <apratap at="" sagebase.org=""> > wrote: >> >> Hi All >> >> I am trying to analyze some mRNA illumina bead level data through bead >> array. Based on the detection p-values(plot attached): if I read it >> right significant # probes >> have no expression which concerns me but I could have easily missed >> some important step. >> >> I am attaching the two box plots 1.) intensity and 2.) probe level >> detection p-values + code >> >> Let me know your opinion. >> >> Thanks! >> -Abhi >> >> beadlevel_data <- readIllumina(dir=dataDir, useImages=F, >> illuminaAnnotation="Humanv4") >> >> #intensity boxplot >> boxplot(beadlevel_data, transFun = logGreenChannelTransform, col = >> "green", >> ylab = expression(log[2](intensity)), las=2, outline=FALSE, >> main= "Array intensities") >> >> #summarize : detection p-val boxplot >> BSData <- summarize(beadlevel_data) >> det = calculateDetection(BSData) >> boxplot(det, main='dset1_mRNA_probe_detection_pvals (Null: there is no >> expression) ', ylab='p-value(detection score)',xaxt='n', >> xlab='individual samples/array' ) >> >> _______________________________________________ >> 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 > > > > > -- > Levi Waldron > Assistant Professor of Biostatistics > City University of New York School of Public Health, Hunter College > 2180 3rd Ave Rm 538 > New York NY 10035-4003 > phone: 212-396-7747 -------------- next part -------------- A non-text attachment was scrubbed... Name: dset1_mRNA_array_singal_to_noise_ratio.png Type: image/png Size: 46883 bytes Desc: not available URL: <https: stat.ethz.ch="" pipermail="" bioconductor="" attachments="" 20140124="" f5a225b1="" attachment.png="">
ADD REPLY
0
Entering edit mode
Just dropping in a gentle reminder if my last update was missed over the weekend. -Abhi On Fri, Jan 24, 2014 at 3:10 PM, Abhishek Pratap <apratap at="" sagebase.org=""> wrote: > Hi Levi > > Thanks for the quick feedback. I am still following up on few of the > things that you recommended. I checked with the collab and these were > not FFPE specimens. Also I did a quick signal to noise ratio > calculation (95% / 5% intensity quantiles ). I am attaching the plot. > May be that indicates something for the Illumina bead array platform > that I am not aware of. To me it looks skewed. > > Cheers! > -Abhi > > On Thu, Jan 23, 2014 at 4:26 PM, Levi Waldron > <levi.waldron at="" hunter.cuny.edu=""> wrote: >> Dear Abhi, >> >> it looks to me like your chips had low hybridization, based both on the low >> percentage of probes detected and on the low interquartile range of >> intensities. Fig. 2 from PMID23136189 (admittedly my publication) shows the >> range of IQRs for samples within several published studies using Illumina >> BeadArrays for FFPE tissues. The normal IQR is around 2, which corresponds >> to around 50-60% present at p<0.05. Your typical IQR is around 1, and I >> can't directly read percent present but it looks around 20-30%. I would be >> concerned, and do some additional checks: 1) if you have technical >> replicates, check them by MA plot and sample-wise correlations, 2) MA plots >> against the median pseudochip, 3) check for detection and high intensity of >> the positive control probes (labelled CY3_HYB and HOUSEKEEPING on Illumina >> chips I've looked at), and 4) run the ffpe::sortedIqrPlot() function to see >> whether IQR or percent present are related to correlation to median >> pseudochip in your experiment. Are you using FFPE specimens? >> >> Sincerely, >> Levi >> >> >> >> On Thu, Jan 23, 2014 at 6:38 PM, Abhishek Pratap <apratap at="" sagebase.org=""> >> wrote: >>> >>> Hi All >>> >>> I am trying to analyze some mRNA illumina bead level data through bead >>> array. Based on the detection p-values(plot attached): if I read it >>> right significant # probes >>> have no expression which concerns me but I could have easily missed >>> some important step. >>> >>> I am attaching the two box plots 1.) intensity and 2.) probe level >>> detection p-values + code >>> >>> Let me know your opinion. >>> >>> Thanks! >>> -Abhi >>> >>> beadlevel_data <- readIllumina(dir=dataDir, useImages=F, >>> illuminaAnnotation="Humanv4") >>> >>> #intensity boxplot >>> boxplot(beadlevel_data, transFun = logGreenChannelTransform, col = >>> "green", >>> ylab = expression(log[2](intensity)), las=2, outline=FALSE, >>> main= "Array intensities") >>> >>> #summarize : detection p-val boxplot >>> BSData <- summarize(beadlevel_data) >>> det = calculateDetection(BSData) >>> boxplot(det, main='dset1_mRNA_probe_detection_pvals (Null: there is no >>> expression) ', ylab='p-value(detection score)',xaxt='n', >>> xlab='individual samples/array' ) >>> >>> _______________________________________________ >>> 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 >> >> >> >> >> -- >> Levi Waldron >> Assistant Professor of Biostatistics >> City University of New York School of Public Health, Hunter College >> 2180 3rd Ave Rm 538 >> New York NY 10035-4003 >> phone: 212-396-7747
ADD REPLY

Login before adding your answer.

Traffic: 865 users visited in the last hour
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