Hello all.

I have array data from two time points for patients with high disease activity.  The disease activity is relatively stable between the time points.

I also have array data from two time points for different patients with stable, low disease activity. 

The hypothesis is that there is differential gene expression between high and low disease activity.  I am wondering how best to incorporate the longitudinal data.  Would this be best used as a replicate?  Or better to model each patient as an explanatory variable?

Thanks

Eric

You have two choices:

1. You could average the two time points for each patients (using avearrays) and then do a simple two group comparison between high and low disease activity, or

2. Analyze as a multi-level experiment as described in the limma User's Guide Section 9.7.

The first option is more conservative, but I would suggest option 2, as it will allow you to test low vs high for the two time points separately.