Goal: Develop metagenomics approaches to assess the functioning of microbial communities in the environment. Q4 Target: Summarize the latest computational approaches to analyze large complex 'omics' datasets to describe microbial community function in environmental samples. Summary: The LANL SFA in Terrestrial Microbial Carbon Cycling aims to inform climate modeling and enable carbon management in terrestrial ecosystems by discovering widespread biological processes that control carbon storage and release in temperate biome soils. To achieve these goals, computational approaches are essential to analyze increasingly complex 'omics' data from microbial communities in environmental samples. The SFA continues to leverage advances in 'omics measurement technology and computation tools to decipher microbial community function in environmental samples. The computational approach is aligned with the recent shift in the SFA's research approach. When the SFA began 10 years ago, computational approaches principally addressed "who is there" with extremely limited insight into function [1, 2]. Accessible 'omics data for complex soil communities were primarily amplicon DNA sequences of taxonomic or functional marker genes obtained by PCR, cloning, and Sanger sequencing. Computational tools for community analysis were nascent. Over the past decade, genome resources (DOE-Integrated Microbial Genomics database) and data acquisition increased 1000-fold, common computational pipelines for many analysis tasks have arisen, and the variety of 'omics' measurements has expanded to routinely include shotgun metagenomics, metatranscriptomics, and versions of metabolomics. The SFA is capitalizing on these advances while also developing new computational techniques to decipher how communities with different patterns of carbon cycling function from the species to ecosystem level. The SFA is establishing for routine use a computational approach that integrates exascale metagenomic computing with multi-scale ecosystem modeling. The computational approach exploits machine learning techniques [3] to reduce the dimensionality of larger 'omics datasets that arise from the SFA's research approach. These techniques yield a subset of features that best predict functional outcomes from microbial community activity [3]. Related computational techniques are used to infer interactions between organisms and metabolic products, yielding specific research targets for mechanistic studies. Comparative genomics [4], exascale computing (NERSC user project) for community metagenome assembly, and metabolic interpolation approaches [5] are being applied to improve the quality and functional interpretation of metatranscriptomic data, which is the most detailed measurement available for community physiology. These approaches are embedded in a larger framework of modeling and simulation of soil carbon cycling. For soil carbon modeling, we are using SOMic 1.0, developed by Cornell collaborators [6]. The SFA is applying SOMic at multiple scales (e.g. [7] to guide the ...