Dear Gordon, excuse me the last mail, I wasn't aware of the fact hatbI sent it to you personally. Dear Gordon, thank you a lot for your fast answer! I attached a schema of the slide-design, also with an example of the 4 replicates, hoping that like this I could explain it sufficiently. For my understanding the replicates are regularly arranged. As example mir 16, which we use as control mir, its 4 replicates can be found in block 5, block 13, block 21 and block 29, always with the same coordinates: column 9, row 3. If I count along the column there are 21 spots between the replicates (NOT counting the replicates), if I count along rows there 127 spots between the replicates, but also this leads to an error message I could not imagine another way to count the distance. I followed your advice, to see if limma tells me the distance using the comand getLayout, but as spacing he tells me not available/missing value $ngrid.r [1] 8 $ngrid.c [1] 4 $nspot.r [1] 11 $nspot.c [1] 16 $ndups [1] 4 $spacing [1] NA attr(,"class") [1] "PrintLayout" > Since limma recognizes the 4 replicates $ndups [1] 4 optimistically I tried spacing 1, but this results in a very weired top table list, where the 4 replicates are shrinked to 2 and other mirs are missing completely, obviously this doesn't work. What would you advice? Thanks a lot, Christine -- Dr. Christine V?llenkle, Ph.D. Research Laboratories-Molecular Cardiology I.R.C.C.S. Policlinico San Donato Via R. Morandi, 30 20097 S. Donato M.se (MI) Italy Phone: +39 02 52774 683 (lab) +39 02 52774 533 (office) Fax: +39 02 52774 666 email: christine.voellenkle at gmail.com

> I attached a schema of the slide-design, also with an example of the 4 > replicates, hoping that like this I could explain it sufficiently. > For my understanding the replicates are regularly arranged. Yes, the replicates on these Exiqon arrays are spaced evenly and at some point I did compute the actual spacing. As you already mentined there are 4 identical sets and the resulting spacing is the number of genes in each of them. Then I discovered that it is a lot more conveniant to simply sort the data by ID - that way you have exactly one requirement: Each spot needs to be replicated the same number of times, while the exact layout, number of genes, blocks etc. becomes irrelavant: MA2 = MA[order(MA$genes$ID), ] dupfit = duplicateCorrelation(MA2, ndups=4, spacing=1) fit <- lmFit(MA2, ndups=4, spacing=1, correlation=dupfit$consensus, design=MA2$design) fit <- eBayes(fit) cu Philipp -- Dr. Philipp Pagel Lehrstuhl f?r Genomorientierte Bioinformatik Technische Universit?t M?nchen Wissenschaftszentrum Weihenstephan 85350 Freising, Germany http://mips.gsf.de/staff/pagel