A Bayesian Approach Inferring Ecostates from Metagenomic Amplicon Data

• Activity Number: 117
• Type: Topic Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): John O'Brien*+
• Companies: Bowdoin College
• Keywords: Metagenomic ; amplicon ; reversible jump ; ecostates

Abstract

Ecostates - assemblages of species working in concert - have become a important focus in understanding the function of microbiomes. However, the statistical approaches for inferring ecostates from metagenomic amplicon data are limited. I propose a new method to model these structures using a multinomial-Dirichlet count likelihood that balances the number of parameters accounting for variation within samples against the complexity of inferred ecostates across samples. I show how this model can be used for case-control studies, and discuss a possible extension to a framework for regression.