Project Summary/ Abstract Metagenomic sequencing provides a means of interrogating the genetic diversity and composi tion of a microbial environment. However, interpretation of these data remains a challenge with current methods for taxonomic profiling involving the 16S rRNA gene, k-mer sequence com position, and nominally single-copy marker genes. None of these techniques is perfect. We propose to develop a new computational method that uses information inherent in the CRISPR Cas prokaryotic immune system: namely, the repeat sequences in each CRISPR array evolve over time and thus contain phylogenetic information. While this technique will only be applica ble to CRISPR-containing microbes, CRISPR can be found in >40% of completely sequenced prokaryotic genomes and has the potential to reveal fine-scale composition differences within that subset. We aim to: 1) evaluate the extent to which CRISPR repeat diversity and taxonomy covary at different taxonomic ranks; 2) build a probabilistic mixture model to infer the most likely community profile from a set of CRISPR repeats; and 3) compare the predictive performance of our inference method to existing methods on both simulated metagenomic data and actual environmental samples.