Development of genome-scale metabolic models and a deep learning model for predicting bacterial genes with significant changes in expression levels

Abstract

Over the past decade, there has been an increasing effort of collaboration between researchers in two seemingly unrelated fields, computer science and biology, which resulted in the emergence of an interdisciplinary field, computational biology. During my time as a master’s student at Systems Biology and Medicine Lab (SBML), I focused on two types of research and development which revolved around the construction of biological networks of industrially important bacteria, particularly Streptomyces rapamycinicus, Corynebacterium glutamicum, and Escherichia coli. In the second chapter, I investigate the application of genome-scale metabolic models (GEMs) for analyzing the metabolic network variations between two strains of Streptomyces rapamycinicus, one of which can overproduce rapamycin as a result of random mutagenesis. Additionally, using GEMs simulations, several metabolic engineering targets for increasing the yield of rapamycin in Streptomyces rapamycinicus and beta-alanine in Corynebacterium glutamicum are proposed. In the third chapter, I will discuss the major update on the development of DeepMGR, a deep learning model for predicting the gene expression levels in Escherichia coli in a given culture media composition. DeepMGR was initiated by my mentor, Mun Su Kwon, in the laboratory during his master’s study, and has been through significant changes regarding the input data, model architecture, and labeling rule to produce a much better prediction.