Thank you for your antwort. I would like to discuss more second question. > > 2. And which mathematical methods for comparison between preprocessing can I > > use? > > One place to look would be the affycomp package. However, you might want to > look at the literature a bit before going too far; some of these comparisons > have been done already with tightly controlled experiments. > I worked before with AffyComp, but the package is not sufficient flexibly. In addition there are limited static information (Assessment summary statistics table) with certain preprocessing methods (MAS, dChip, RMA). simply said, I want more than median SD, R2, correlation, etc.. and I look for meaningful Literature and source about the problem. I am pleased about new ideas and discussions, which can help me. Thanks in advance, Mohammad Sean Davis <sdavis2 at="" mail.nih.gov=""> schrieb am 04.01.06 22:25:16: > > > > > On 1/4/06 2:57 PM, "Mohammad Esad-Djou" <shahrgol at="" web.de=""> wrote: > > > > > Hello, > > > > I wants to compare expression measures, which was computed with MAS and RMA. > > > > But I noticed that the expression-values are very different. > > > > e.g.: Expt1_R1.CEL: Gen: 1007_s_at > > > > RMA: 8,06941 > > > > MAS: 236,16853028 > > > > 1. How can I interpret this difference? > > MAS is not in log2 units. RMA is in log2 units. > > > 2. And which mathematical methods for comparison between preprocessing can I > > use? > > One place to look would be the affycomp package. However, you might want to > look at the literature a bit before going too far; some of these comparisons > have been done already with tightly controlled experiments. > > Hope that was what you were asking.... > Sean > >

On 1/9/06 2:47 PM, "Mohammad Esad-Djou" <shahrgol at="" web.de=""> wrote: > Thank you for your antwort. > > I would like to discuss more second question. > >>> 2. And which mathematical methods for comparison between preprocessing can I >>> use? >> >> One place to look would be the affycomp package. However, you might want to >> look at the literature a bit before going too far; some of these comparisons >> have been done already with tightly controlled experiments. >> > > I worked before with AffyComp, but the package is not sufficient flexibly. > > In addition there are limited static information (Assessment summary > statistics table) with certain preprocessing methods (MAS, dChip, RMA). > simply said, I want more than median SD, R2, correlation, etc.. and I look for > meaningful Literature and source about the problem. Since affycomp is open source, you can certainly extend their methods with whatever measures you like. However, I think you need to know what measures you want to use and what your goals are. Sean