Firstly, I'd like to congratulate James and Gordon on doing an excellent job with limmaGUI! My question is a simple one. The arrays I work with tend to have duplicates on each slide. However, they are not adjacent nor equally spaced and so when setting the spacing of the duplicates in limmaGUI (which appears to be compulsory) I just make it up. I presume this makes the correlation between duplicates meaningless but that linear model fits are OK. Comments? Cheers Peter -- Dr Peter Baker, Statistician (Bioinformatics/Genetics), CSIRO Mathematical & Information Sciences, Queensland Bioscience Precinct 306 Carmody Road, St Lucia Qld 4067. Australia. Email: <peter.baker@csiro.au> WWW: http://www.cmis.csiro.au/Peter.Baker/ Phone:+61 7 3214 2210 Fax:+61 7 3214 2900

Hi Peter, At 04:58 PM 15/10/2003, Peter Baker (CMIS, St Lucia) wrote: >Firstly, I'd like to congratulate James and Gordon on doing an >excellent job with limmaGUI! > >My question is a simple one. > >The arrays I work with tend to have duplicates on each slide. However, >they are not adjacent nor equally spaced and so when setting the >spacing of the duplicates in limmaGUI (which appears to be compulsory) >I just make it up. > >I presume this makes the correlation between duplicates meaningless >but that linear model fits are OK. The short answer is that you cannot use limma or limmaGUI to handle duplicate spots unless they are equally spaced. See https://stat.ethz.ch/pipermail/bioconductor/2003-August/002224.html Telling limma that you have duplicates with a given spacing when you don't will make everything meaningless, including the linear model fit, because you will be causing limma to average the results from spots which actually correspond to different genes. Gordon >Comments? > >Cheers >Peter