Expanding the limits of Scale and sensitivity in microbial genomics

Abstract

Sequencing of microbial genomes has enabled new understanding of human health and disease. Certain microbes can support human health through the microbiota, helping to train our immune system or supply essential nutrients. In other cases, microbes may be pathogenic, overwhelming the immune system and causing infection. Low-cost and accessible DNA sequencing has allowed us to learn important information about microbial systems - we can identify what microbes are members of our microbiota and how they change with disease, as well as how pathogenic microbes evolve and acquire resistance to antibiotics. While the cost of sequencing has decreased and allowed for widespread use, studies are now limited by sample acquisition and preparation. In particular, microbial sample preparation has challenges at the limits of sensitivity (low signal to noise ratio) and at the limits of scale (large sample size). In this thesis, I developed methods to address both of these challenges and applied the techniques to study questions in basic biology and in clinical medicine. First, I developed a procedure to sample and sequence the lung microbiome in mouse models, where high background of mammalian DNA in lung samples poses a serious challenge for sequencing preparation. Along with my collaborator, I used this procedure to investigate the microbiome in a murine model of lung cancer. Second, I developed a platform for high-throughput sequencing preparation of bacteria at the scale of thousands of samples, with a 100-fold less cost per sample. I prepared and sequenced 3000 antibiotic-resistance bacteria from a clinical trial studying the role of decolonization procedures. This work provides new insights about microbes in the context of health and disease, and the methods developed here can make samples newly accessible for sequencing at the limits of scale or sensitivity.