Regularized Estimation of Microbial Ecological Networks

• Activity Number: 117
• Type: Topic Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Zachary Kurtz*+ and Richard A. Bonneau and Christian L. Mueller and Emily Miraldi and Martin Blaser and Eric J. Alm
• Companies: and New York University and New York University and New York University and NYU Lagone Medical Center and MIT
• Keywords: microbiome ; computational biology ; compositional data analysis

Abstract

High-throughput sequencing has enabled quantification of microbial communities, the microbiome, across diverse ecosystems. The identification of microbial mechanisms requires new statistical tools as microbiome abundance data presents several challenges:

Foremost, the abundances of microbial populations are compositional, that is, normalized to the total observed counts. Therefore, microbial components are not independent and traditional statistical metrics (e.g., correlation) can lead to spurious results.

Microbial sequencing-based studies typically measure thousands of populations on tens to hundreds of samples. This suggests that any abundance-based model of microbial interactions is likely under-determined and additional information is required to accurately infer networks.

In this setting, we developed a new method that integrates sparse estimation of the ecological network interactions and compositionally-robust measures of proportionality from microbiome population data.

We benchmarked our method by simulating community structures with a given network topology. We inferred networks from ecological datasets to generate testable hypotheses about microbial interactions.