Lets wait on this one until I have a chance to analyze the data more completely. --Naomi At 12:37 AM 6/27/2005, Michael Kirk wrote: >Quoting Naomi Altman <naomi at="" stat.psu.edu="">: > > > Kirk, > > You have missed my point about the 50 control spots. These are all the > > SAME oligo. > > The correlation here is induced by the array, not by the RNA > concentration, > > hybridization efficiency, etc. > >OK, if it's the same oligo then that does indicate a spot, or hybridization >effect. Not suprising I guess. Out of curiosity - did you quantify the size >of the effect? I.e. was the inter replicate within slide variation of >significant >size relative to whatever variation may be expected between treatments? > >Michael > > > > > The reason that the two color analysis is supposed to be more efficient > > than 1 channel is the positive correlation between the > > errors for the 2 channels on the same spot. If the channels are > > uncorrelated, then there is no spot effect and using the differences is no > > better than using the 2 channels. > > > > The single channel analysis can be used, providing that you use a linear > > mixed model that includes a random effect for array (and, in the case of > > multiple spots per gene) for > > spot(array). > > > > --Naomi > > > > > > At 09:20 PM 6/26/2005, Michael Kirk wrote: > > >While I agree that it is probably a bad idea to use single channel > > >analysis on two colour arrays, some of the arguments presented here > > >are a little troubling. > > > > > >The observation that the intra slide correlation is 0.8 doesn't, to my > > >mind, show anything unless it is high relative to inter slide > > >correlation. Regardless of what treatments are applied to the samples, > > >all mouse (say) samples would be expected to have roughly similar > > >(array wise) expression profiles. This is partly a reflection of the > > >fact that many genes may not vary between treatments and different > > >probes will have different hybridization efficiencies (i.e. some spots > > >will always have low intensities and some high). > > > > > >Secondly, IF the single channel intensities were in fact highly > > >accurate, then it is the two colour analysis that would be inefficient > > >(in terms of number of arrays required). The two colour idea is > > >essentially to overcome noise, particularly noise due to variation in > > >the printed spots between slides (i.e. the chemical/physical > > >properties of a spot for a given gene may vary between slides). In > > >this case the variation is assumed to affect each hybridized sample > > >similarly (multiplicatively) and by taking the ratio this variation is > > >removed. A fine idea, but it does leave us with less information than > > >if the slide quality was sufficient for this to to be unnecessary. > > > >From the two colour analysis of a single slide we have a set of > > >ratios, which may then be compared between slides. From the single > > >channel analysis of a two colour hybridization we have two sets of > > >measurements, which also may be compared between slides. > > > > > >With two colour analysis, only three samples can be compared using two > > >slides, whereas if the single channel analysis was justified (and note > > >I am not say it is, only discussing the arguments given against it), > > >then four samples can be compared. > > > > > >Michael > > > > > > > Wolfgang, > > > > > > > > Naomi is refering to what I call the "intraspot" correlation, see for > > > > example the intraspotCorrelation() function in the limma package, and > > > it is > > > > critically important. The correlation isn't a bad thing, nor is it > > > > restricted to poor quality arrays. Rather it means that contrasts > > > estimated > > > > within a spot are highly accurate. It is what makes the two- colour > > > > technology intrinsically more accurate than one channel technology, > other > > > > things being equal. See > http://www.statsci.org/smyth/pubs/ISI2005-116.pdf > > > > for some discussion. > > > > > > > > Basically, you're saying that if the arrays are very high quality, > you can > > > > get away with an inefficient analysis. Why not do it properly and > get the > > > > full benefit of the high quality arrays? My experience is that high > > > quality > > > > Agilent arrays can beat affy for accuracy if treated properly. > > > > > > > > Gordon > > > > > > > > >Date: Thu, 23 Jun 2005 15:29:38 +0100 (BST) > > > > >From: "Wolfgang Huber" <huber at="" ebi.ac.uk=""> > > > > >Subject: Re: [BioC] Agilent Arrays > > > > >To: "Naomi Altman" <naomi at="" stat.psu.edu=""> > > > > >Cc: bioconductor at stat.math.ethz.ch > > > > > > > > > >Hi Naomi, > > > > > > > > > >and why is that important? Also, what is the within gene correlation > > > > >between green foreground of array 1 and green foreground of array 2? > > > > > > > > > >Bw > > > > > Wolfgang > > > > > > > > > ><quote who="Naomi Altman"> > > > > > > I am working with Agilent arrays on which we have spotted many > > > replicates > > > > > > of the control spots. > > > > > > The within gene correlation between red and green forground is > > > about 0.8 > > > > > > for the unnormalized data - i.e. pretty high! > > > > > > > > > > > > --Naomi > > > > > >[snip] > > > > > >_______________________________________________ > > >Bioconductor mailing list > > >Bioconductor at stat.math.ethz.ch > > >https://stat.ethz.ch/mailman/listinfo/bioconductor > > > > Naomi S. Altman 814-865-3791 (voice) > > Associate Professor > > Bioinformatics Consulting Center > > Dept. of Statistics 814-863-7114 (fax) > > Penn State University 814-865-1348 (Statistics) > > University Park, PA 16802-2111 > > > > > >_______________________________________________ >Bioconductor mailing list >Bioconductor at stat.math.ethz.ch >https://stat.ethz.ch/mailman/listinfo/bioconductor Naomi S. Altman 814-865-3791 (voice) Associate Professor Bioinformatics Consulting Center Dept. of Statistics 814-863-7114 (fax) Penn State University 814-865-1348 (Statistics) University Park, PA 16802-2111