Hello Ben, Our collaborators have been taking microdissected samples and amplifying them. We find that the distribution of probe intensities in a probe set is greatly affected; the signal from 5' probes is much decreased. However if the amplifications are done carefully, the amplified samples can be compared to each other. We find the RNA degradation plots in affy a good way to assess comparability. I don't think amplified samples could be compared with un-amplified samples, if that was your aim. Regards Mark Original Message: Date: Wed, 1 Dec 2004 14:49:08 -0500 From: "Wittner, Ben" <wittner.ben@mgh.harvard.edu> Subject: [BioC] correcting for 3' bias in highly amplified RNA samples and Affyme trix chips To: "Bioconductor List (bioconductor@stat.math.ethz.ch)" <bioconductor@stat.math.ethz.ch> Message-ID: <c6b4e960236fd311b4e00008c7f406e31404b6f8@phsexch11.mgh.harvard.edu> Content-Type: text/plain Is there some way in Bioconductor to correct for 3' bias in Highly amplified RNA samples for Affymetrix chips. I know that Affy now has a 3' biased human chip, but I have human data that was done with another chip and cannot be re-done. I also have mouse data and I don't think there is a 3' biased mouse chip from Affy. I've heard of people doing such a correction (not necessarily using BioC or R), but I can't find any mention of it in the affy documentation or in the Bioconductor mailing list archives. Thanks. -Ben

Thanks Mark. I think all your comments below are right. Nevertheless, I expect that the signal to noise ratio would be improved by de-emphasizing the results from the probes at the 5' end because presumably their signal-to-noise is worse than that of the probes at the 3' end. -Ben Date: Thu, 2 Dec 2004 09:40:44 -0500 From: "Reimers, Mark (NIH/NCI)" <reimersm@mail.nih.gov> Subject: [BioC] RE: correcting for 3' bias in highly amplified RNA samples To: "'bioconductor@stat.math.ethz.ch'" <bioconductor@stat.math.ethz.ch> Message-ID: <16A0583FB1644E4DB8C0A0265028B6FD9306EB@nihexchange13.nih.gov> Content-Type: text/plain Hello Ben, Our collaborators have been taking microdissected samples and amplifying them. We find that the distribution of probe intensities in a probe set is greatly affected; the signal from 5' probes is much decreased. However if the amplifications are done carefully, the amplified samples can be compared to each other. We find the RNA degradation plots in affy a good way to assess comparability. I don't think amplified samples could be compared with un-amplified samples, if that was your aim. Regards Mark Original Message: Date: Wed, 1 Dec 2004 14:49:08 -0500 From: "Wittner, Ben" <wittner.ben@mgh.harvard.edu> Subject: [BioC] correcting for 3' bias in highly amplified RNA samples and Affyme trix chips To: "Bioconductor List (bioconductor@stat.math.ethz.ch)" <bioconductor@stat.math.ethz.ch> Message-ID: <c6b4e960236fd311b4e00008c7f406e31404b6f8@phsexch11.mgh.harvard.edu> Content-Type: text/plain Is there some way in Bioconductor to correct for 3' bias in Highly amplified RNA samples for Affymetrix chips. I know that Affy now has a 3' biased human chip, but I have human data that was done with another chip and cannot be re-done. I also have mouse data and I don't think there is a 3' biased mouse chip from Affy. I've heard of people doing such a correction (not necessarily using BioC or R), but I can't find any mention of it in the affy documentation or in the Bioconductor mailing list archives. Thanks. -Ben