Hi, it's hard to say what causing this, but if you see it in several samples at the same location, then my immediate thought is that you reference signal may carry it. Are you using the average of the pool of all samples as a reference or how do you calculate it? How many samples to you have in your reference pool? CN polymorphic regions that are frequent enough in your population could cause this, but then it should be a real biological signal, which you say it isn't.
Are you using the full or the default GenomeWideSNP_6 CDF? Affymetrix removed several CN loci from the former to make the latter - CN loci that they found to be poor for CN analysis. This could also be a reason though those loci should be scattered fairly randomly along the genome. You could also check if there is a difference between the signals from SNPs and CN loci. If there is, that would indicate that there is some artifacts on the arrays. Also, are you really sure you are using the correct annotation data? For instance, if you use the full CDF to generate the data, but only the default for extracting genome locations (assuming the same ordering of row indices), such weird things may show up. If you plot your data using the ChromosomeExplorer, this should be taken care of automatically. Also, do some QC plots using ArrayExplorer; there might be spatial artifacts, although it sounds unlikely. Sorry, not much help, but at least some directions for troubleshooting. /Henrik On Jul 21, 5:34 pm, Matt Wilkerson <mdwilk...@gmail.com> wrote: > Hello, > > I have detected what I think is an array quality issue and wanted to get > others' opinions about this phenomena. > > I observed this issue on chromosome views of CN from SNP6 arrays. It looks > like a "smearing" effect where CN has irregular values and a range of large > negative numbers to zero within specific regions. The regions at which this > happens are identical in affected samples and occur on basically all > chromosomes. This smearing is not cancer DNA segment loss, where probes > belonging to a segment have similar CN values. In a group of about 70 > arrays, 1/3 of the arrays have this issue and the others have expected > segments of discrete amplifications/deletions. I have compared specimen, > technical, and array characteristics to try to find a batch or quality > issues, but the effect appears so far to be randomly occuring. > > I put an example > at:http://www.unc.edu/~mwilkers/artifact.png<http://www.unc.edu/%7Emwilkers/artifact.png> > In the plot, the points are probes. Axes are base position and log2 median > centered copy number. The lines are segments overlaid. The colors are not > important. > > I don't think this is an aroma issue - I detect the phenomena using > apt-copynumber-workflow also. The only affymetrix summary option that > associated with the artifact samples was "allele summarization mean". The > artifact arrays had lower values. > Also, I have used aroma successfully with 250K_Sty arrays often and never > seen this phenomena. > > My question: > Has anyone seen this phenomena before? > Does anyone have an explanation or suggestion? > > Thank you, > > Matt Wilkerson > Lineberger Comprehensive Cancer Center > University of North Carolina at Chapel Hill -- 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/
