不良研究所

It is well understood that breast cancer is an extremely heterogeneous disease, and that valuable clinical information is reflected in the transcriptional profiles of tumors. Traditional methodologies to analyze the profiles of breast carcinoma have revealed the presence of distinct subtypes among tumors, and an association of several important biological processes with these subtypes. However, our work to date suggests that other key processes may operate orthogonally to these groups, and are instead activated across a subset of tumors that belong to the various subtypes. Determining which tumors express these secondary processes is difficult, since the molecular signal of such processes is often overshadowed by the dominating responses associated with the primary subtypes. We have designed a framework to address this challenge. Our methodology relies on the linear ordering of a panel of patients over specific gene sets, rather than classification of samples into discrete groups using entire transcriptional profiles. Crucially, our framework is able to assess the relative activation level of individual biological pathways within a tumor. This, in turn, allows breast tumors to be defined by the molecular processes that affect their phenotype. Our framework contains the necessary tools to assess the activation level of numerous molecular pathways in breast tumors, and further reveals how these pathways are related to one another.