Question about the context-specific regulon generation

[xiaoyonf]

Dear ARACNe-AP authors,

Thank you for developing such a great tool in gene network construction. I am trying to reconstruct a gene network using a list of epigenetic regulators and the TCGA breast cancer RNA-seq data. It is well known that gene expression profiles are very different between molecular subtypes of breast cancer, e.g., hormone receptor-positive vs. negative tumors, which are expected to have different gene regulatory network centered on the epigenetic regulators I am focusing on. My question is if I only look for the gene regulatory network in a particular molecular subtype (i.e., hormone receptor-positive breast cancer), should I only use the hormone receptor-positive subset of the TCGA breast cancer RNA-seq data, instead of the whole dataset including both hormone receptor-positive and negative, to construct the network? In other words, if I construct the network using the whole TCGA breast dataset, whether the molecular subtypes with underlying different regulatory mechanisms of epigenetic regulation, will bias the construction of the final inferred network?

Thank you in advance for your reply.

Best,

Xiaoyong Fu
Baylor College of Medicine
Houston, TX