Hi Mohamed: It is quite arbitrary to determine the threshold for detection p value when filtering out non-expressed probes. But a cutoff of 0.1 will normally remove 30-40% of all the probes in an experiment from my experience. This cutoff will remove those probes which have intensities not greater than 90% of negative controls. Hope this helps. Cheers, Wei Mohamed lajnef wrote: > Dear Wei, > > Thanks for your messages! > > I found the Detection scores colomns (Detection Pval-which > estimates the probability of a probe being detected above the > background level) in my data. > to detect the probes which do not express, I must remove the probes > that have Pvalue>0.05 in the detection scores colomns before carrying > out differential expression analysis???????????? > > > thank you so much for your help > > Cheers > Mohamed > > 2009/3/31 Wei Shi <shi@wehi.edu.au <mailto:shi@wehi.edu.au="">> > > Hi Mohamed: > > Nonexpressed probes are not those probes which have missing > values. We do not know probes with missing values are expressing > or not. You can use "Detection" scores to try to find those probes > which do not express in all of your arrays and then filter them > out before carrying out differential expression analysis to see if > you can get some DE genes. You should be able to find "Detection" > columns in your expression profiles. > > > Cheers, > Wei > > Mohamed lajnef wrote: >> Dear Wei & Sean, >> >> Sean: result of topTbale >> > gene1 >> ID logFC AveExpr t P.Value >> adj.P.Val B >> 9623 7570474 -0.11659435 4.991279 -3.826220 0.0002074251 >> 0.9999788 0.4950223 >> 43599 5550736 -0.18627341 8.316864 -3.800538 0.0002275799 >> 0.9999788 0.4162686 >> 931 4570630 -0.10900890 4.834441 -3.776569 0.0002480571 >> 0.9999788 0.3431365 >> 20081 5560544 -0.22781231 8.704546 -3.770575 0.0002534446 >> 0.9999788 0.3249047 >> 33970 6280711 -0.08123219 5.013244 -3.521244 0.0006066949 >> 0.9999788 -0.4135212 >> 19419 4010192 -0.09110624 4.903243 -3.516622 0.0006163498 >> 0.9999788 -0.4268336 >> 37533 3290356 0.08945260 4.876439 3.512031 0.0006260841 >> 0.9999788 -0.4400442 >> 10727 1260035 -0.14783999 5.784915 -3.506596 0.0006377940 >> 0.9999788 -0.4556641 >> 9921 2260719 -0.09260158 4.927686 -3.474897 0.0007103085 >> 0.9999788 -0.5463862 >> 17444 6180360 0.08237090 4.932764 3.474807 0.0007105246 >> 0.9999788 -0.5466424 >> >> >> >> Wei: 1) i used beadarray methods( quantile) to normalise the log2 >> transoformed data >> 2) i dont have a missing values in my data! >> >> Thank you again for help >> >> >> >> >> >> >> 2009/3/30 Wei Shi <shi@wehi.edu.au <mailto:shi@wehi.edu.au="">> >> >> Hi Mohamed: >> >> I think you can try two things: >> >> (1) Try alternative normalization methods; >> (2) Remove probes which are not expressing in all your arrays. >> >> Hope this helps. >> >> Cheers, >> Wei >> >> Mohamed lajnef wrote: >> >> Dear All, >> >> I'am using limma package to extract the genes >> diffrentially expressed by 3 >> treatment,my database includes >> >> 48803 genes (rows) and 120 colones (40 replicates per >> treatment),my program >> is as follows: >> >> donne<-exprs(BSData.quantile) >> groups<-as.factor(c(rep("Tem",40),rep("EarlyO",40),rep("LateO",40))) >> design<-model.matrix(~0+groups) >> colnames(design)=levels(groups) >> fit<-lmFit(donne,design) >> cont.matrix<-makeContrasts(TemvsEarlyO=Tem- EarlyO,TemvsLateO=Tem-LateO,EarlyOvsLateO=EarlyO-LateO, >> levels=design) >> fit2<-contrasts.fit(fit, cont.matrix) >> ebfit<-eBayes(fit2) >> gene1<-topTable(ebfit, coef=1) >> gene2<-topTable(ebfit, coef=2) >> gene3<-topTable(ebfit, coef=3) >> >> results<-decideTests(ebfit) # I have a matrix that >> contains only 0 because >> >> in the ligne 26 results <- new("TestResults", >> sign(tstat) * (p < p.value)) >> of decideTests program the condition (p<p.value) is="">> always false in my >> case!! >> something is wrong in my program, but I dont know where?? >> >> Any help would be appreciated. >> >> Regards >> >> [[alternative HTML version deleted]] >> >> _______________________________________________ >> Bioconductor mailing list >> Bioconductor@stat.math.ethz.ch >> <mailto:bioconductor@stat.math.ethz.ch> >> https://stat.ethz.ch/mailman/listinfo/bioconductor >> Search the archives: >> http://news.gmane.org/gmane.science.biology.informatics.conductor >> >> >> > [[alternative HTML version deleted]]