Input: G (gene network), M (expression matrix), y (clinical outcome vector), T (local fdr threshold), C (confounder matrix) Output: Collection of DC genes: S Standardize each row of M Local Moran’s I Matrix I = Local _ Moran _ I(G, M) For each node i in  G do ti = t statistic from generalized linear model y~Ii + C End for Fit {ti}i = 1, …, p to mixture model using local fdr to find {lfdri}i = 1, …, p S = {i : lfdri ≤ T} Return S |