Hi all, I am new to using bioconductor. I was wondering if there was a way to extract the p-values that are used to make the P/M/A calls when one runs the "mas5calls" function in affy. Secondly, I am wondering if someone has already come up with a solution for the bigger problem I am facing. I have data from 6 affy chips (3 expt and 3 control). what I want to be able to do is get rid of all the probe sets that are called absent in both sets. I have seen a report that uses the Fisher test on pvalues (that are used to make the A/M/P calls) and come up with a cumulative pvalue (which follows a chi square distribution). Then "If the combined p-value for a given probe set is greater than a cutoff (0.05) for both control and experiment (i.e. called "absent" in both samples), then the probe set is filtered out". This is exactly what I want to do with my data. Any sugggestions on how to do this? Thanks Much Kartik

Hi Kartik, I use a simpler method to remove absent probesets. If a probeset has at least 1 'present' call or at least 2 'marginal' calls, then I keep it , otherwise I throw it out. This may not be as sophisticated as what you want to do, but it will remove almost all of the probesets you want to remove. Here's a piece of code I use to convert the P/M/A calls to numbers: calls.eset <- mas5calls(rawdata) num.pres <- apply(exprs(calls.eset), 1, function(x) { x[x == "A"] <- 0 x[x == "P"] <- 1 x[x == "M"] <- 0.3 x <- as.numeric(x) return(sum(x)) }) You can change the valuation of the M calls, but I use 0.3 to distinguish 2 Marginal from 1 Present. Then, you can subset an exprSet object using whatever numerical cutoff you want: eset.pres <- eset.all[num.pres > 0.5 , ] Cheers, Jenny At 07:28 PM 8/17/2006, Kartik Pappu wrote: >Hi all, > >I am new to using bioconductor. I was wondering if there was a way to >extract the p-values that are used to make the P/M/A calls when one >runs the "mas5calls" function in affy. > >Secondly, I am wondering if someone has already come up with a >solution for the bigger problem I am facing. I have data from 6 affy >chips (3 expt and 3 control). what I want to be able to do is get rid >of all the probe sets that are called absent in both sets. I have >seen a report that uses the Fisher test on pvalues (that are used to >make the A/M/P calls) and come up with a cumulative pvalue (which >follows a chi square distribution). Then "If the combined p-value for >a given probe set is greater than a cutoff (0.05) for both control and >experiment (i.e. called "absent" in both samples), then the probe set >is filtered out". > >This is exactly what I want to do with my data. Any sugggestions on >how to do this? > >Thanks Much >Kartik > >_______________________________________________ >Bioconductor mailing list >Bioconductor at stat.math.ethz.ch >https://stat.ethz.ch/mailman/listinfo/bioconductor >Search the archives: >http://news.gmane.org/gmane.science.biology.informatics.conductor Jenny Drnevich, Ph.D. Functional Genomics Bioinformatics Specialist W.M. Keck Center for Comparative and Functional Genomics Roy J. Carver Biotechnology Center University of Illinois, Urbana-Champaign 330 ERML 1201 W. Gregory Dr. Urbana, IL 61801 USA ph: 217-244-7355 fax: 217-265-5066 e-mail: drnevich at uiuc.edu

Hi Kartik, At the R prompt, try: ?mas5calls The help page shows that mas5calls returns an exprSet object with p-values in the se.exprs slot If you really want to do present call filtering as you describe, just off the top of my head, you might try: (1) Extracting pvalues from the object resulting from mas5calls, e.g. mas5calls.out: pvals<-mas5calls.out at se.exprs (2) Create a function to determine the Fisher pvals that accepts a vector of the call p-values as input ... look at GeneMeta and the vignette "Meta analysis for microarrays" for instructions/code for how this might be accomplished. I call it "yourmeta" below. Maybe someone else can help you out here. (3) Apply the function to each row of "raw" pvals to generate a vector of cumulative pvalues for each probeset: fisherpvals<-apply(pvals,1,yourmeta) (4) Filter by rows for which the cumulative pvalues are less than 0.05. filtered.data<-exprs(alldata)[fisherpvals < 0.05,] Assuming your exprSet is named "alldata" Is that what you wanted? I don't know much about GeneMeta, it might contain something like "yourmeta" already. I wish I could be more helpful right now, but this should get you started. Best, Duane Hassane -----Original Message----- From: bioconductor-bounces@stat.math.ethz.ch [mailto:bioconductor-bounces at stat.math.ethz.ch]On Behalf Of Kartik Pappu Sent: Thursday, August 17, 2006 7:29 PM To: bioconductor at stat.math.ethz.ch Subject: [BioC] Extracting pvals for PA calls in affy Hi all, I am new to using bioconductor. I was wondering if there was a way to extract the p-values that are used to make the P/M/A calls when one runs the "mas5calls" function in affy. Secondly, I am wondering if someone has already come up with a solution for the bigger problem I am facing. I have data from 6 affy chips (3 expt and 3 control). what I want to be able to do is get rid of all the probe sets that are called absent in both sets. I have seen a report that uses the Fisher test on pvalues (that are used to make the A/M/P calls) and come up with a cumulative pvalue (which follows a chi square distribution). Then "If the combined p-value for a given probe set is greater than a cutoff (0.05) for both control and experiment (i.e. called "absent" in both samples), then the probe set is filtered out". This is exactly what I want to do with my data. Any sugggestions on how to do this? Thanks Much Kartik _______________________________________________ Bioconductor mailing list Bioconductor at stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/bioconductor Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor -----Original Message----- From: bioconductor-bounces@stat.math.ethz.ch [mailto:bioconductor-bounces at stat.math.ethz.ch]On Behalf Of Kartik Pappu Sent: Thursday, August 17, 2006 7:29 PM To: bioconductor at stat.math.ethz.ch Subject: [BioC] Extracting pvals for PA calls in affy Hi all, I am new to using bioconductor. I was wondering if there was a way to extract the p-values that are used to make the P/M/A calls when one runs the "mas5calls" function in affy. Secondly, I am wondering if someone has already come up with a solution for the bigger problem I am facing. I have data from 6 affy chips (3 expt and 3 control). what I want to be able to do is get rid of all the probe sets that are called absent in both sets. I have seen a report that uses the Fisher test on pvalues (that are used to make the A/M/P calls) and come up with a cumulative pvalue (which follows a chi square distribution). Then "If the combined p-value for a given probe set is greater than a cutoff (0.05) for both control and experiment (i.e. called "absent" in both samples), then the probe set is filtered out". This is exactly what I want to do with my data. Any sugggestions on how to do this? Thanks Much Kartik _______________________________________________ Bioconductor mailing list Bioconductor at stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/bioconductor Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor