R help - Sep 2010 - Multivariate linear model versus univariate linear model and interaction

Hi,

I have a quick question regarding the merits of modeling data using a
multivariate linear model versus a univariate linear model with an
interaction term. Here's the data set. Say I have gene expression from two
tissues liver and kidney and I'm interested in identifying differentially
expressed genes in liver only, in kidney only or in both tissues. I could
fit the following model:

expression ~ condition*tissue

but I could also theoretically treat the expression levels as a two
dimensional response and fit it as:

Expression ~ condition

where Expression is now a matrix although in the latter case I'm not sure
how to test for differential expression in only one tissue, any ideas?

Thanks,

~Jimmie

	[[alternative HTML version deleted]]