Hello,

I'm having trouble figuring out how to design my analysis.

I have 10 pairs of DNA from 10 subjects. The first DNA extraction takes place at study entry and then the second DNA extraction takes place 3 months later.

Over the 3 months,

1. Some subjects exhibited an increase in blood iron concentration.
2. Some subjects exhibited no change in blood iron concentration,
3. Some subjects exhibited decrease in blood iron concentration.

So ultimately I want to see how CHANGE in blood iron concentration (predictor) effects expression of specefic genes (response variable).

I was thinking a paired sample analysis would work with Limma as this would allow me to treat each pair as a block, however I run into the issue of CHANGE in blood concentration as my predictor (a continuous phenotype).

The only targets frame I can think of is the one below, but I have a feeling it will not accomplish the goal of determining how CHANGE in blood concentration effects the expression of genes.

Any advice on the correct way of setting up my analysis? Thanks in advance

ID pair conc time_point
1   1   20  baseline
2   1   30  follow-up
3   2   20  baseline
4   2   20  follow-up
5   3   30  baseline
6   3   10  follow-up

#conc goes up for pair 1, conc stays the same for pair 2, conc goes down for pair 3 

# include your problematic code here with any corresponding output 
# please also include the results of running the following in an R session 

sessionInfo( )

Just put (log) blood iron concentration and pair in the design matrix. limma will work with differences automatically -- there's no need for you to compute them explicitly. Any gene which is DE for the blood iron coefficient has changes correlated with blood iron changes.