A Non-Negative Tensor Factorization Approach to Deconvolute Microenvironment in Breast Cancer

Abstract

An in-depth understanding of epithelial breast cell response to growth-promoting ligands is required to elucidate how signals from the microenvironment affect cell-intrinsic regulatory networks and their resultant cellular phenotypes, such as cell growth, progression, and differentiation. Understanding the cellular response to these signals is particularly important in understanding the mechanisms of breast cancer initiation and progression. There is increasing evidence that aberrant epigenetic marks are present in cells of the breast tumor microenvironment and are known to affect these cellular processes. However, the mechanisms by which epigenetic microenvironment signals influence these cellular phenotypes are complex and currently not well established. To deconvolute the complexity of the epigenetic microenvironment signals in breast cancer, we developed a novel tensor-based correlation method: HOCMO (Higher-Order Correlation Model), applying to proteomics time series data to reveal the four-way regulatory dynamics among signaling proteins, histones, and growth-promoting ligands across multiple time points in the breast epithelial cells. HOCMO reveals two functional modules and the onset of specific protein-histone signatures in response to growth ligands contributing to distinct cellular phenotypes indicative of breast cancer initiation and progression. We evaluate robustness of our tensor model against baseline method TensorLy and achieved slight improvement in terms of reconstruction error and execution time. HOCMO is a data independent self-supervised learning method with superior interpretability that can capture the strength of complex interactions such as inter- and intra-pathway cellular signaling networks in any diseases or biological systems.