I have used java based ARACNe-AP to find interaction partners of genes involved in metabolic pathway of my interest. I have data (65 samples) from tumor and matched samples. I want to understand what are novel interactions present in tumor environment. When I run the program, I get a list of "Regulators", "Targets" and associated "mutual information score" and "p-value". I selected a gene of interest and made a cut off of 0.6 MI. Still there are hundreds if interaction. When I use the same gene in normal samples and with the same MI cutt-off of 0.6, I get 5-8 times for partners. I feel I am doing something wrong. Could you anyone share their experience using the ARACNe-AP framework?