Algorithms for E. coli genome engineering

Abstract

Author summary: Lamba red recombineering is one of methods of performing genome engineering. However, this method of genome editing is not very specific and efficient and is highly dependent on the genomic regions that are targeted (integration sites). In this project we explored ways of identifying what makes a site well suited for lambda red genome engineering. We wanted to explore whether we can eventually predict the "goodness" of an integration site using an algorithm. Our initial approach to the problem was to write an algorithm based on some characteristics that we felt would be key to determining the goodness of a site. Choosing to initially focus on specificity of the integrations, we used experimental approaches to evaluate whether our algorithm had any predictive powers for specificity. Upon failing, we revised our plan to generate a dataset of ~150 sites and their integration data (whether integration was successful, specific and efficient at that site). We used this dataset to explore correlations between the specificity data and characteristics we thought might affect the specificity of sites. The most promising characteristics appeared to be the uniqueness of the genomic site (as determined by BLAST) and the existence of Repetitive Extragenic Palindrome (REP) sites at the site of integration. Section I of this thesis sets up the problem, section II talks about the initial approach we took to the problem and section III discusses our modified approach -- which formed the bulk of this thesis project. Section I and III are the most relevant to understand the project, while Section II gives more content to the project in addition to detailed insight to what approaches did not work.