UC Davis

Computational methods play an irreplaceable role for optimization of biological organisms in the era of high-resolution omics, genetic engineering, and high-performance computing. A general overview of computational methods for optimization of biological organisms is presented in Chapter 1 with a focus on three main challenges relating to data scarcity and heterogeneity, model interpretability, and the large number of factors that can affect an organisms’ phenotype. Recent advances are discussed in Chapter 2 with a forward-looking view on the application of computational methods for microbiome-based diet and health optimization. In Chapter 3, existing computational methods are applied for microbiome-based diet optimization in irritable bowel syndrome (IBS). The integrated data analysis results argue that there are two types of patients distinguishable by their fecal samples, those with high colonic methane and SCFA production, who will respond well on a low-FODMAP diet, and all others, who would benefit a dietary supplementation containing butyrate and propionate, as well as probiotics with SCFA-producing bacteria, such as lactobacillus. In Chapter 4, a novel artificial neural network (ANN) architecture called genetic neural network (GNN) is presented that captures the dependencies and non-linear dynamics that exist in gene networks into the GNN architecture. The results argue for 40% more accuracy of GNNs compared to several common ANNs in predicting genome-wide gene expression given gene knockouts and master regulator perturbations in bacterium E. coli. In Chapter 5, a novel group testing method called algorithmic lifestyle optimization (ALO) is presented for rapid identification of effective lifestyle interventions in individuals. ALO is robust to noise, data size and data heterogeneity, is between 58.9% and 68.4% more efficient compared to standard elimination diet for identification of food items that exacerbate IBS symptoms and allergic reactions, and better than alternative state of the art group testing method for this application. The conclusions and future directions are discussed at the end of each chapter and summarized in the final chapter. Chapters 2, 3 and 4 are published (1–3).