A Bayesian method for detecting pairwise associations in compositional data
Open Access
- 15 November 2017
- journal article
- research article
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 13 (11) , e1005852
Abstract
Compositional data consist of vectors of proportions normalized to a constant sum from a basis of unobserved counts. The sum constraint makes inference on correlations between unconstrained features challenging due to the information loss from normalization. However, such correlations are of long-standing interest in fields including ecology. We propose a novel Bayesian framework (BAnOCC: Bayesian Analysis of Compositional Covariance) to estimate a sparse precision matrix through a LASSO prior. The resulting posterior, generated by MCMC sampling, allows uncertainty quantification of any function of the precision matrix, including the correlation matrix. We also use a first-order Taylor expansion to approximate the transformation from the unobserved counts to the composition in order to investigate what characteristics of the unobserved counts can make the correlations more or less difficult to infer. On simulated datasets, we show that BAnOCC infers the true network as well as previous methods while offering the advantage of posterior inference. Larger and more realistic simulated datasets further showed that BAnOCC performs well as measured by type I and type II error rates. Finally, we apply BAnOCC to a microbial ecology dataset from the Human Microbiome Project, which in addition to reproducing established ecological results revealed unique, competition-based roles for Proteobacteria in multiple distinct habitats. Data from many fields are available primarily in the form of proportions, also referred to as compositions, which impose mathematical constraints on identifying interactions among components in the underlying systems. In particular, correlations cannot be calculated directly from proportions or from count data that give rise to them. Methods that work around this difficulty generally do so by imposing strong assumptions about the distribution of underlying data or associated correlations, and these in turn often prevent quantifying uncertainty in the resulting estimates of correlation. We developed a statistical model (BAnOCC: Bayesian Analysis of Compositional Covariance) that both estimates correlations between counts or proportions and provides a posterior distribution for each correlation that quantifies how uncertain the estimate is. BAnOCC does well at controlling the number of false positives in simulated data and can be practically applied to a wide range of proportional data types.Keywords
Funding Information
- National Institute of Diabetes and Digestive and Kidney Diseases (U54DE023798)
- National Science Foundation (DBI-1053486)
- Army Research Office (US) (W911NF-11-1-0473)
- National Science Foundation (US) (ATD-1042785)
- National Institutes of Health (US) (R01HG005220)
This publication has 31 references indexed in Scilit:
- A Guide to Enterotypes across the Human Body: Meta-Analysis of Microbial Community Structures in Human Microbiome DatasetsPLOS Computational Biology, 2013
- Intestinal Inflammation Targets Cancer-Inducing Activity of the MicrobiotaScience, 2012
- Inferring Correlation Networks from Genomic Survey DataPLOS Computational Biology, 2012
- Microbial Co-occurrence Relationships in the Human MicrobiomePLOS Computational Biology, 2012
- Structure, function and diversity of the healthy human microbiomeNature, 2012
- Using network analysis to explore co-occurrence patterns in soil microbial communitiesThe ISME Journal, 2011
- Oral multispecies biofilm development and the key role of cell–cell distanceNature Reviews Microbiology, 2010
- The Pangenome Structure ofEscherichia coli: Comparative Genomic Analysis ofE. coliCommensal and Pathogenic IsolatesJournal of Bacteriology, 2008
- Characterization and modeling of the Haemophilus influenzae core and supragenomes based on the complete genomic sequences of Rd and 12 clinical nontypeable strainsGenome Biology, 2007
- Bayesian Variable Selection in Linear RegressionJournal of the American Statistical Association, 1988