I am having trouble inputting my experimental design within MeV. It is a 2 x 2 factorial design (Old/Young; Control/Treated), using two color arrays, where all comparisons were made, and all dye-swapped. Here is a schematic view of my design :
Old Control<->Old Treated
^ ^ ^ ^
! \ / !
! x !
! / \ !
v v v v
Young Control<->Young Treated
Every two-color array is thus a particular combination of these four factor (eg, young.control vs old.treated)
For now, I've been doing a normalization step outside MeV (or Midas), and the data I input in Mev consists of normalized log-ratios (one column per array).
I want to use the Limma or two-factor ANOVA module to get a differentially expressed gene list, but I'm stuck because each of my column are not one particular factorial combination (eg, old contol) but one particular comparisons between two factorial combinations (eg, young.control vs old.treated).
I must have been missing something in the way MeV handle two color arrays, and what my input data should be like.
Any help on this matter to get me through a 2 x 2 factorial analysis with two-color arrays in MeV would be great.
MeV expects samples to be in one factor combination. For one-color arrays this is simple. For two-color arrays these samples are typically run against a common reference sample rather than a different factorial combination (If I understand your experimental design).
Thanks and good luck,
Thanks Dan for your support !
So basically nobody knows how to handle this kind of experimental design in two-color arrays with MeV ? Or is it simply that it is not possible to handle it ?
I'd be rather surprised that this really complete software cannot handle the very experimental design that statisticians advocate in technical reviews !
Any help is welcome.