Andrew E Teschendorff
Cancer System-omics: a statistical mechanics perspective
One of the greatest challenges facing the cancer omics community is the inherent complexity of cellular biology. While the generation of large-scale multi-omic data, including notably single-cell data, promises to advance our understanding of cellular biology, we are currently still lacking theoretical and computational frameworks capable of taming the underlying complexity. In this talk, I will broadly argue that we need to develop a statistical mechanical framework for cellular biology, which can help us predict emergent properties at the cellular population level. I will advocate a framework based on the concept of signalling entropy, which I demonstrate is able to predict important features of the cellular differentiation landscape, of cancer and drug resistance. I will further describe an application of statistical mechanics to the problem of risk prediction and early detection of cancer. Specifically, in a proof of principle study, we show how applying our framework to DNA methylation data from cytologically normal cervical smear samples can predict the prospective risk of cervical neoplasia up to 3-years prior to diagnosis. I will end by describing our efforts to generalize this risk prediction framework to other women specific cancers.