Probabilistic Reconstruction and Comparative Systems Biology of Microbial Metabolism
With the number of sequenced microbial species soon to be in the tens of thousands, we are in a unique position to investigate microbial function, ecology, and evolution on a large scale. The path from genotype to phenotypic predictions can be understood as a three step process: 1) genome annotation, 2) molecular network reconstruction, and 3) simulation of network behavior. Here I explore these three aspects of systems biology focusing on microbial metabolic networks. I start by describing GLOBUS, a novel probabilistic method for the genome-wide annotation of metabolic genes. GLOBUS annotations integrate multiple types of heterogeneous data including network context correlations and species phenotypic properties. Second, I describe the curation of the genome-scale metabolic network for the malaria parasite, Plasmodium falciparum. The reconstruction allowed the identification of many potential essential genes and the validation of one of them as a possible drug target for malaria treatment; the model is also a platform to contextualize multiple types of molecular data. Finally, I present a comparative analysis of phenotypic properties across hundreds of bacterial species that spans 4 billion years of evolution. The implications of the analysis for understanding microbial biodiversity and evolution are discussed.