To construct the gene dependency networks, we utilized conditional mutual information (CMI) and permutation tests to identify pairs of genes that were dependent on each other during phenotypic changes. We then employed the GeneRank algorithm [16,17] to rank the importance of each gene in AD patients based on their expression levels and dependencies. Using these rankings, we identified nutrients that could target crucial genes as potential interventions for preventing or treating AD. Figure 1 depicts our analysis workflow.
