[Genome] GNF Atlas 2 Median Ratios...

[Michael Mayhew]

Hello,

    I am attempting to build a distribution out of the 'expression 
values' for a set of genes (say of N=200) for all tissues but one (say 
all tissues but brain).

    I am using the GNF Atlas 2 median ratios to build this distribution. 
So, what I end up with is a vector of points with one point for each 
tissue except brain for each gene (so 200*78 points).

    I have noticed a large number of duplicate median ratios in the 
vector of points I create to build my distribution. (in one case ~11000 
/ ~16000) I have already mapped all RefSeq IDs to Entrez Gene IDs so 
that each Entrez ID has its own set of median ratios and so that the 
duplications are not coming from multiple median ratio vectors for 
multiple isoforms of the same gene.

    I have spoken to microarray users in my group and they have all 
advised me that log-ratios (in general) should be very minimally 
duplicated (~1% was one person's estimate).

    Is there anything inherent in the median ratios that would result in 
this kind of pattern (perhaps the number of significant digits)? Is 
there something I could be missing in building my vector of points?

    Thank you very much for your time and consideration in this matter.

Michael