[BioC] Limma and Affymetrix ?

Stan Smiley swsmiley at genetics.utah.edu
Wed Feb 11 23:39:31 MET 2004

```I have been looking for some examples of limma with Affymetrix data. There
is a simple example in the documentation, but not a case study as with the
dye swap experiments, and I am confused as to how to proceed with using
limma on my data.

I need to eventually analyze a 2^4 factorial experiment, but I'd like to
start with a smaller example so I can get the idea first. I know what
contrasts are of interest to me, but I am unclear on how to use them with
limma.

ie. For a 2^3 factorial experiment:
3 Factors A,B,C ; each at two levels, high and low.

Treat		Factors
ments	A	B	C
c	low	low	low
a	high	low	low
b	low	high	low
c	low	low	high
ab	high	high	low
ac	high	low	high
bc    low	high	high
abc	high	high	high

would be:

Single: For A,B or C alone
2x interactions: AB, AC and BC
3x interaction : ABC

With corresponding contrasts of:

4A   = -c + a - b + ab - c + ac - bc + abc
4B   = -c - a + b + ab - c - ac + bc + abc
4C   = -c - a - b - ab + c + ac + bc + abc

4AB  = +c - a - b + ab + c - ac - bc + abc
4AC  = +c - a + b - ab - c + ac - bc + abc
4BC  = +c + a - b - ab - c - ac + bc + abc

4ABC = -c + a + b - ab + c - ac - bc + abc

I'm unsure how to proceed in getting significant genes
from each contrast.

Any help and/or references to examples etc. would be greatly
appreciated!