Langille, M.G.I. et al. Nat. Biotechnol. doi:10.1038/nbt.2676 (25 August 2013).

Researchers performing microbial community sampling ask the question: “Who is there?” Single marker–gene sequencing has proven cost-effective for microbe identification, but it cannot reveal what the microbes in a community are capable of doing. To bridge the gap, Langille et al. serve up PICRUSt (phylogenetic investigation of communities by reconstruction of unobserved states), a computational tool that predicts functions from marker-gene profiling data. PICRUSt first extracts gene content from every microbe with a genome sequence in a reference phylogenetic tree; then it predicts the gene content and corresponding traits of related unsequenced microbes using an adaptation of ancestral state reconstruction methods. From 16S ribosomal DNA marker sequences, it can reproduce a large share of the functional information found by shotgun metagenomic sequencing in the Human Microbiome Project and other data sets.