Hi Simon, > Hi Heidi, > hopefully you can respond to the last email with this one. Should I write > you on the regular forum? I'm a bit unclear about the etiquette here. > In general, writing to the forum is preferable. 1) any email to [BioC] has a greater chance of getting noticed in my inbox, 2) it serves as a Q&A repository, and 3) you may get a useful reply from other people. > In any case, I have 5 plates of data in a 48.48 biomark format, with > different layouts on each plate with regards to the samples, but the same > genes are on every plate. > > after following your last set of suggestions, I can get the sample ID's > associated with the correct genes. I'm having some basic trouble > generating some of the plots though, as I'm getting out of range errors. > This is due to the fact that we've got a lot of groups (roughly 30 groups > or so). This leads to incorrect assignation of margin size on the plots, a > fairly typical problem in R. My question is, do you have some general > guidelines to solve this? e.g., a series of iterative steps to try when a > plot is not working correctly due to wrong margin settings etc. > The short answer would be no. What you can possibly try though is for example: - setting the margins yourself before plotting, using e.g. par(mar=c(3,2,2,1); plotXXX() - plotting directly to a file, where you can increase the height and width dimensions - opening a plotting device such as X11 or quartz, likewise with height and width specified, before calling the plot command More specifically, what's the dimensions of your qPCRset? I.e. how many sample groups do you have. And what plots are failing? If this is likely to be commonly occurring as qPCR platforms increase in size, I should look into it. > The other question is combining the 5 plates I have into one object. > You've covered this to some extent in your guide, but I was wondering > whether or not having exactly the same distribution of samples per plate > affects the merging. For example, some of our groups have 5 samples per > group, while as others have 4, and this varies from plate to plate > I assume you're using cbind(), in which case it shouldn't matter. As long as the genes are in the same order, it's okay. With 5 48.48 plates you should just end up with 48 features (rows) x 240 columns. The content of the columns can vary in any way, as long as you indicate it correctly during the calls to functions that require the data to be grouped by samples. Hope this helps, and I apologise for being so tardy with my replies. As you've undoubtedly noticed, HTqPCT wasn't originally planned for data in the BioMark format, and I'm just trying to catch up with it. If you have any analysis steps/issues that you think ought to be explained in the vignette, then please let me know. \Heidi > I hope I'm clear! > > thanks again for all your efforts, you guys do a terrific job in helping > many people. > > best > > s > > Simon Melov Ph.D. > Associate Professor & > Director of Genomics > Buck Institute for Research on Aging > 8001 Redwood Blvd > Novato, CA 94945 > > Office: 415 209 2068 > Cell: 415 827 4979 > Fax: 415 209 9920 > > > > > > >