Variational Autoencoding Tissue Response to Microenvironment Perturbation

摘要

This work applies deep variational autoencoder learning architecture to study multi-cellular growth characteristicsof human mammary epithelial cells in response to diverse microenvironment perturbations. Our approachintroduces a novel method of visualizing learned feature spaces of trained variational autoencoding models thatenables visualization of principal features in two dimensions. We find that unsupervised learned features moreclosely associate with expert annotation of cell colony organization than biologically-inspired hand-crafted features,demonstrating the utility of deep learning systems to meaningfully characterize features of multi-cellular growthcharacteristics in a fully unsupervised and data-driven manner.