[BioC] M vs M plots- dye effect in dyeswap exp

Naomi Altman naomi at stat.psu.edu
Wed Feb 8 17:07:45 CET 2006


As usual I hit "send" too quickly.  I take back the comment about the 
plots of R vs R and G vs G.

--Naomi

At 10:48 AM 2/8/2006, zarzabal wrote:
>Thanks for replying.
>I was assured that these pairs are dyeswaps.
>Yes, normalization was performed and the scatter plots look good 
>after normalization.
>
>As Sean suggested I went back and reviewed all my plots. The density 
>plots for the kidney arrays look very similar but the bimodal peak 
>is more defined in the kidney arrays. For the liver arrays the plots 
>are less similar across arrays and only a couple still have the 
>bimodal peak after normalization. The individual scatter plots look 
>pretty good after normalization, but the dyeswap slides do not look 
>like mirror images of each other-shouldn't they?. Some almost look 
>identical. The background images all seem to favor the green dye 
>more than red and there is definitely a spatial artifact from the 
>missing spots in the center.
>
>I also reviewed the mean/median intensities of both the raw 
>intensities (red and green) and the normalized log ratio values. The 
>values for each channel looks pretty comparable across arrays. Of 
>course I am very new at this so I am not sure what sizable 
>differences are acceptable or not. I am curious though, shouldn't my 
>median values for each dyeswap pair be close? There was a very big 
>difference on a few of the pairs.
>
>As for the clustering suggestion, I do not have all the data, (print 
>batch, RNA extraction) needed.
>
>I am really stumped on this. I am not a biologist so please excuse 
>me for my lack of knowledge as far as the experiments are concerned. 
>I went back and reviewed some data with excellent looking dyeswaps 
>and the only difference that I know of is that the RNA was company 
>grade, not in-house. Could this just be an issue of RNA quality? 
>Anymore suggestions?
>
>Thanks again
>LZ
>
>Naomi Altman wrote:
>
>>If these pairs are dye-swaps, then the M-values should be 
>>negatively correlated.  In this case, you have a problem (assuming 
>>you normalized the data first).
>>
>>If they are not dye-swaps, then perhaps they are OK.  You cannot 
>>pay too much attention to the smooths outside the main body of the data.
>>
>>--Naomi
>>
>>At 11:19 AM 2/6/2006, lzarzabal at aol.com wrote:
>>
>>
>>>Hello All,
>>>
>>>I was wondering if anyone had any suggestions on quality issues 
>>>with dyeswap slides.  We ran 2 exp's with dye swap slides and some 
>>>of them looked questionable.  I have attached the M vs M 
>>>plots.  Also several of the density plots are bimodal.  We had 
>>>missing spots that were missed by the print tips, which I removed. 
>>>I read in previous BioC emails that if there is an x pattern on 
>>>the MvsM plot that it could be due to degradation of the 
>>>dyes.  The experimenters have assured me that the dyes are not 
>>>expired and are from a very reputable company.  Does anyone have 
>>>any ideas, are the slides usable or need to be redone?  We are 
>>>using in-house RNA.
>>>
>>>Thanks
>>>LZ
>>>
>>>
>>
>>_______________________________________________
>>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
Dept. of Statistics                              814-863-7114 (fax)
Penn State University                         814-865-1348 (Statistics)
University Park, PA 16802-2111



More information about the Bioconductor mailing list