Hi Steven. On Mon, Jul 26, 2010 at 9:00 PM, Steven Bosinger <steven.bosin...@gmail.com> wrote: > Hi, > > I'm new to aroma and bioC in general, so these are probably a very > straightforward questions: > > I am using aroma to get QC on some Human Affymetrix Gene arrays. > > 1. To keep it consistent with previous analyses using RMA pre- > processing (BKG subtraction, quantile normalization and median polish > summarization), can I use the medpolish function instead of RmaPlm?
Not sure what implementation you used before, but the RMA summarization step in the oligo package uses median polish. RmaPlm does not do the low-level calculation itself, by rely on existing code/package for this. By default it is using affyPLM/preprocessCore. You can tell RmaPlm to use that of oligo as: plm <- RmaPlm(..., flavor="oligo"); There are some more comments in help(RmaPlm). Note that even if two different implementations/software say they are using median polish, they may not be numerically reproducible. How median polish is started, how many iteration it runs etc may give you different results. If I remember correctly, it is also known for not always converging, i.e. it can oscillate between two results. Note that median polish and rlm (robust linear modelling) are both estimator for the same log-additive probe-level model, i.e. they try to estimate the same parameters but in different ways. Some people/software documentations are sloppy and say they use median polish, but in reality they might actually have used rlm. I would recommend to use rlm, if possible. You can always run both variants and see how much the results differ. > > 2. I read in the forum that NUSE plots aren't available when you > summarize using medpolish, is this the case? Good catch. Could you provide a link where you found that? In order to calculate NUSE (Normalized Unscaled Standard Errors), you need standard deviations of the parameter estimates. The median polish estimator [see help(oligo::basicRMA)] does *not* give/return standard deviation/errors of the parameter estimates. Internally, we fix the stddev to 1 (one), so if you try to calculate NUSE, you'll get nothing useful or even an error. > > 3. Is there a vignette/pdf file similar to BioC that lists all the > available functions for aroma? The website is the main source of documentation. > > 4. How can I export the RMA pre-processed data matrix to another 3rd > party software? A good start is probably to use extractDataFrame() and write its content in a format you like. See the how-to page 'Extract probeset summaries (chip effects) as a data frame': http://aroma-project.org/howtos/extractDataFrame > > 5. Is there a function for MvA plots? It's not clear to me *what* you want to plot, but basically: plotMvsA(cf, reference=cfR); where 'cf' and 'cfR' are two AffymetrixCelFile:s, e.g. cf <- getFile(cs, 1); cfR <- getAverageFile(cs); You can do the same by replacing the AffymetrixCelSet 'cs' with an ChipEffectSet 'ces'. > > 6. How do I format plots? ie alter range, color etc The same way as you usual do - what have you tried and what didn't work? > > Sorry for these newbie questions... No worries. Though, next time, please try to post one question/topic per message, and try to be more precise in what you are asking/have tried. Then it is easier to help and quicker to reply to. Cheers, Henrik > > Steve. > > -- > When reporting problems on aroma.affymetrix, make sure 1) to run the latest > version of the package, 2) to report the output of sessionInfo() and > traceback(), and 3) to post a complete code example. > > > You received this message because you are subscribed to the Google Groups > "aroma.affymetrix" group with website http://www.aroma-project.org/. > To post to this group, send email to aroma-affymetrix@googlegroups.com > To unsubscribe and other options, go to http://www.aroma-project.org/forum/ > -- When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example. You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/. To post to this group, send email to aroma-affymetrix@googlegroups.com To unsubscribe and other options, go to http://www.aroma-project.org/forum/
