Dear Jianhong, Maybe someone else can tell you how much different preprocessing algorithms may effect the outcome. However, you can also test it yourself: Simply preprocess your data with the Affymetrix Power Tools (APT) first using RMA and second using PLIER, and then apply MIDAS to both data and compare the results. Regards Christian On 5/24/12 3:52 PM, Ou, Jianhong wrote: > Dear Christian, > > Thanks for your reply. > > I thought event the algorithms are different, the predicted alternative splicing events should be similar. So I compared the probeset_id list of different expressed which determined by different methods. I also tried only compare the top 100 predicted events, the number of overlapping probeset_ids is small. > > As you experience, is it normal? > > Yours sincerely, > > Jianhong Ou > > jianhong.ou at umassmed.edu > > > On May 23, 2012, at 1:45 PM, cstrato wrote: > >> Dear Jianhong, >> >> You do not describe what you are comparing: >> >> You say that you tried oligo, xps, aroma.affymetrix, and also APT(Affymetrix Power Tools). Then you say that only the results of MiDAS from APT and results of FIRMA from aroma.affymetrix have about 40% overlap. All the others have less than 10% overlap. >> What do you mean with less than 10% overlap??? >> What did you compare??? >> >> Please note that MIDAS was developed by Affymetrix as alternative to ANOVA to measure differences between exon level signal... >> The Affymetrix "exon_alt_transcript_analysis_whitepaper.pdf" says: >> The basic idea is the following: >> ? We use PLIER to generate a robust estimate... >> >> It seems that you are comparing apples and oranges, since you seem to compare MIDAS (based on PLIER) to RMA!? >> >> If you want to compare the different packages you need to compare the results obtained from e.g. RMA and not from MIDAS, which most packages did not implement (although xps has implemented FIRMA). >> >> With respect to APT vs xps please read the vignette "APTvsXPS.pdf" where the reasons for the slight differences between the RMA results obtained with APT or xps are explained. >> >> Regards >> Christian >> _._._._._._._._._._._._._._._._._._ >> C.h.r.i.s.t.i.a.n S.t.r.a.t.o.w.a >> V.i.e.n.n.a A.u.s.t.r.i.a >> e.m.a.i.l: cstrato at aon.at >> _._._._._._._._._._._._._._._._._._ >> >> >> >> >> On 5/23/12 2:55 AM, Ou, Jianhong wrote: >>> Dear lists, >>> >>> I am doing some analysis of HuEx-1_0-st-v2 chip. I tried R package oligo, xps, aroma.affymetrix. I also tried Affymetrix power tools, JETTA and Atlanalyze. After that I feel very confused about my results. Only the results of MiDAS from APT and results of FIRMA from aroma.affymetrix have about 40% overlap. All the others less than 10% overlap. Is that normal? >>> >>> The code for MiDAS is from paper: Exon array data analysis using Affymetrix power tools and R statistical software, Helen E. Lockstone, Brief Bioinform (2011), doi: 10.1093/bib/bbq086 >>> >>> The code for FIRMA is from webpage: http://www.aroma- project.org/vignettes/FIRMA-HumanExonArrayAnalysis and followed with log2 transform and do analysis with LIMMA. >>> >>> The code for JETTA is from webpage: http://gluegrant1.stanford.edu/~junhee/JETTA/instructions.html#rex >>> >>> All the parameters are followed the sample codes. >>> >>> Please, help me. >>> >>> Yours sincerely, >>> >>> Jianhong Ou >>> >>> jianhong.ou at umassmed.edu<mailto:jianhong.ou at="" umassmed.edu=""> >>> >>> >>> >>> [[alternative HTML version deleted]] >>> >>> _______________________________________________ >>> 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 >>> > > >