Abstract A14: Modeling drug combination sensitivity with Hopfield networks and transcriptomics data

The Hopfield neural network model is one of the simplest models able to mathematically implement Waddington’s interpretation of normal and anomalous cell phenotypes as dynamical attractors of epigenetic landscapes. Here, we propose a computational approach based on Hopfield’s associative memories that integrate gene expression data and gene interactome networks in (1) a model representing the dynamics and control of disease progression in multiple myeloma (MM), and (2) a model describing the control of angiogenesis. The MM model is built using single-cell RNA-seq data from bone marrow aspirates of MM patients as well as patients diagnosed with monoclonal gammopathy of undetermined significance (MGUS) and smoldering multiple myeloma (SMM), two medical conditions that often progress to full MM. We identify different clusters of MGUS, SMM, and MM cells, map them to Hopfield associative memory patterns, and model the dynamics of transition between the different patterns. The model is then used to identify combinations of genes whose simultaneous inhibition is associated with delayed disease progression. In the angiogenesis control model, we use single-cell RNA-seq data to predict specific combinations of targets for inhibition that could induce a cellular transition from tip-like to stalk-like endothelial cells, inhibiting the formation of capillary sprouts in blood vessels. The model generates novel hypotheses for combinations that could complement standard VEGF inhibitors and lead to a more efficient control of angiogenesis in cancer. Citation Format: Carlo Piermarocchi, Sergii Domanskyi, Alex Hakansson, Giovanni Paternostro. Modeling drug combination sensitivity with Hopfield networks and transcriptomics data [abstract]. In: Proceedings of the AACR Special Conference on the Evolving Landscape of Cancer Modeling; 2020 Mar 2-5; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2020;80(11 Suppl):Abstract nr A14.