Capturing the most wanted taxa through cross-sample correlations

The Human Microbiome Project (HMP) identified the 16S rRNA gene sequences of 'most wanted' taxa—prevalent in the healthy human microbiota but distant from previously known sequences. Since 2012, few of the corresponding genomes have been isolated and sequenced, and only through advanced isolation techniques. We demonstrate that the genomes of the most wanted taxa can be identified computationally through their correlation in abundance across multiple public metagenomic data sets. We link over 200 most wanted sequences with nearly complete genome sequences, including half of the taxa identified as high-priority targets by the HMP. The genomes we identify have strong similarity to genomes reconstructed through expensive isolation techniques, and provide a more complete functional characterization of these organisms than can be extrapolated from their 16S rRNA gene. We also provide insights into the function of organisms for which 16S rRNA gene signatures were recently reported to be associated with health and host genetic factors.