A Bioinformatics Platform for Studying Metabolic Functions of Unculturable Microbes

The processes of natural ecosystems are strongly affected by the lifestyles of their most numerous inhabitants, the bacterial and archaeal prokaryotes. Examples of phenomena that affect local ecosystems and the biosphere at large include cycling of carbon and nitrogen, and both positive and negative interactions with plants and animals. Recent advances in molecular techniques such as 16S rDNA sequencing have revealed that natural ecosystems are dominated by bacteria which are not culturable and hence remain largely unknown. The growing wealth of completely sequenced microbial genomes does not directly illuminate these natural ecosystems. Direct amplification or cloning from environmental samples of genes involved in metabolic processes of interest does allow us to investigate molecular processes at the genetic level among these poorly known systems. A major informatics challenge is to integrate data from the sequencing of multiple different genes from a mixture of unknown organisms and assemble a model of which genes co-occur within individual species. We propose to develop an informatics-based solution to this problem based on concordance of gene phylogenies and sequence attributes such as codon bias and oligonucleotide word frequencies. The system will take anonymous DNA sequences from multiple loci and estimate which ones come from the same species. A successful solution will enable a new level of sophistication in the way we can study the functional genomics of the largely unknown, unculturable microbes that dominate terrestrial, subterranean, and aquatic ecosystems.