[BioC] limma: positive correlation for dyeswap pairs

[Naomi Altman]

I have 4 arrays in technical replicate dyeswap pairs.  Using 
duplicateCorrelation (LIMMA) I got a positive correlation for the consensus 
correlation and I cannot understand how this is possible.  There is really 
no evidence of dye-bias.   But 13% of the noncontrol spots have correlation 
higher than 0.9.  I have done extensive QC on these arrays, and I am happy 
with the quality of the data.

I have attached a plot of one of the dyeswap pairs.  The left panel is M 
array 1 vs. M array 2.  The highly correlated spots are indicated in 
red.  The right panel is A array 1 vs A array 2.  The histogram is the 
value of A for the highly correlated spots on array 1.

The correlation is even higher if I pair the wrong set of arrays together 
i.e. the correct blocking is 1,1,2,2 but if I put in 1,2,2,1 for the 
blocks, 20% of the spots have correlation higher than 0.9.

If anyone has any ideas, I would be happy to hear them.  The only thing I 
can think of is dye bias.


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