Microbiome. An ecosystem composed of microscopic organisms. Although unseen by the naked eye, these communities can have powerful impacts on their hosts and surrounding environments. Yet, we are just beginning to crack the surface as to who these tiny critters are, how they are surviving, and what their overarching purpose is in the tree of life. This thesis presents software methods developed to improve understanding of these communities by leveraging the advent of high-throughput sequencing and viewing each ecosystem holistically, motivated by the intention of improving upon methods for gut microbiome analysis in concussion recovery.

We dive into three computational tools developed for improvement of understanding the diversity within microbial communities: Emu for taxonomic community profiling, Rhea for structural variant detection, and Kiwi for P4 phage satellite detection. Each of these algorithms was designed with the view of the microbiome as a single evolving entity, rather than a sum of unique individuals. Viewing microbiomes through this lens and incorporating computer science theories in expectation-maximization, graph motifs extraction, and sub-string minimizers allowed us to develop software for each of these concepts that showed improvement upon existing methods.