Checking the scale of the cytokine data

[aarmey]

We should normalize the cytokine data to prevent any batch/sample problems. First, it'd be good to check that there aren't patients who are just low in all of the cytokines. Then, the cytokine data should probably be normalized across cytokines by normalizing according to the geometric mean (i.e. the cytokine geometric mean should be constant for each patient).