PROJECT SUMMARY The CDC SEARCH for Diabetes in Youth study found that Type I Diabetes (T1D) incidence increased by 1.8% each year between 2002-2012, and Type II Diabetes (T2D) increased by 4.8%. The Environmental Determinants of Diabetes in the Young (TEDDY) study has attributed a substantial burden of T1D and T2D to environmental contaminants. Due to the increasing prevalence of diabetes and metabolic diseases (especially among youth), computational models for developmental pancreatic toxicity are needed. Understanding how multiple factors such as chemical structure, gene expression and target tissue cytotoxicity integrate and impact pancreatic health is vital. However, an integrated analysis of multiple factors at multiple scales poses great challenges due to the inherent complexity, high-dimensionality, uncertainty, and heterogeneity. Multilayer networks have emerged as a novel methodology in network science that combines multiple networks, called “layers”, into one mathematical object. Multilayer networks are able to represent multiple factors across multi- scales for a rigorous computational analysis of their interactions. Thereby, uncovering novel relations between key factors on a multi-scale. The overarching goal of this research is to create multilayer network models by which we can predict the magnitude and mechanisms of pancreatic developmental toxicity based on chemical structure in a zebrafish (Danio rerio) model. Aim 1 will build a Quantitative Structure-Activity Relationship (QSAR) model to predict mechanisms of toxicity resulting from pharmacological and toxicological exposures in the developing pancreas. The goal of Aim 1 is to utilize a multilayer network and topological clustering model to predict the relationship between exposures and pancreatic developmental toxicity based on chemical structure. Aim 2 will utilize multi-scale modeling to create an Adverse Outcome Pathway (AOP) using molecular, structural, and pathological criteria for pancreatic developmental toxicity. The goal of Aim 2 is to characterize the processes by which exposures may disrupt pancreas development and early diabetic pathogenesis. We will develop a rigorous predictive model that can be used to better inform a priori testing and expected outcomes of small molecules in the context of pancreatic developmental diseases, and we will construct a framework to connect peroxisome proliferator-activated receptor (PPAR) modulation (pharmacological & toxicological) with aberrant pancreatic development and early function.

PROJECT NARRATIVE Pediatric diabetes has been steadily increasing in recent decades by far greater rates than can be predicted by diet and lifestyle alone. This research utilizes computational modeling to characterize how exposures to pharmacological or toxicological agents during development can impair pancreatic organogenesis and potentially predispose individuals to diabetes later in the lifecourse. Here, we employ a rigorous multilayer network and topological clustering approach, rather than simple regression, to build predictive quantitative structure-activity relationship (QSAR) models and adverse outcome pathways (AOPs) with broader toxicological translation.