Efficacy of ARACNE algorithm for inferring canine B-cell lymphoma gene regulatory network (GRN)

Lymphoma is the most frequent hematopoietic cancer in dogs. Canine B-cell lymphoma has been proposed as an ideal model of human non-Hodgkin’s lymphoma (NHL). Critical genes playing important roles in the cancer progression can be detected using the reconstruction and analysis of gene regulatory network (GRN). GRNs are inferred using various computational algorithms, where ARACNE is on the most important and efficient algorithms. Here, we evaluated the efficacy of ARACNE to reconstruct canine B-cell lymphoma GRN via different computational analyses. Hence, the gene expression profile of GSE43664 was downloaded from GEO database and differentially expressed genes were extracted using statistical analysis. Then, significant genes were subjected to reconstruct GRN using ARACNE algorithm. Our findings indicated that ARACNE inferred a logic biological network with 387 nodes (genes) and 845 edges (interactions). The inferred network followed a biological scale-free pattern, because many nodes had low numbers of interactions and a few nodes were highly connected. Additionally, node degree distribution showed a decreasing linear pattern. Although the network had 80 connected components, most nodes (71.5 %) contributed in a sub-network implying a strong biological network.