Hello, we work with the affy package and we try to compare the results from bioconductor with the results from Affymetrix. In the function expresso you can choose the avgdiff as summary method. We noticed that the function avgdiff uses the median of the probes, but the Affymetrix 4 Algorithmen uses a trimmed mean. In the source code there is a function called avdiff, which calculates the expression value in the same manner as Affymetrix. Why does this function is not implemented in the expresso? I think to take the median, as the function avgdiff does, is very different to the method proposed fby Affymetrix. Tanja Hölter Dipl. Statistikerin Tanja Hölter Institut für Medizinische Informatik, Biometrie und Epidemiologie Universitätsklinikum Essen D-45122 Essen Tel:(49)201-723-4654 FAX: (49)201-723-5933

it is different. using express (another function) you can use the predefined function avgdiff. you can do this using expresso as well but its bit more complicated. in any case, you wont be able to reproduce Affy's avg diff using the included functions becuase, for example, we havent implemented the MAS 4.0 background correction. writting your own is relatively easy. im sure bioconducotor would gladly welcome a contributed function that reproduces MAS 4.0 Avg diff measure. thanks, -rafael On Thu, 5 Dec 2002, [iso-8859-1] "Hölter, Tanja" wrote: > Hello, we work with the affy package and we try to compare the results > from bioconductor with the results from Affymetrix. > In the function expresso you can choose the avgdiff as summary method. > We noticed that the function avgdiff uses the median of the probes, but > the Affymetrix 4 Algorithmen uses a trimmed mean. In the source code > there is a function called avdiff, which calculates the expression value > in the same manner as Affymetrix. Why does this function is not > implemented in the expresso? I think to take the median, as the function > avgdiff does, is very different to the method proposed fby Affymetrix. > > Tanja Hölter > > > > Dipl. Statistikerin Tanja Hölter > Institut für Medizinische Informatik, > Biometrie und Epidemiologie > Universitätsklinikum Essen > D-45122 Essen > Tel:(49)201-723-4654 > FAX: (49)201-723-5933 > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > http://www.stat.math.ethz.ch/mailman/listinfo/bioconductor >