BioMiCo: a supervised Bayesian model for inference of microbial community structure

Abstract

BackgroundMicrobiome samples often represent mixtures of communities, where each community is composed of overlapping assemblages of species. Such mixtures are complex, the number of species is huge and abundance information for many species is often sparse. Classical methods have a limited value for identifying complex features within such data.ResultsHere, we describe a novel hierarchical model for Bayesian inference of microbial communities (BioMiCo). The model takes abundance data derived from environmental DNA, and models the composition of each sample by a two-level hierarchy of mixture distributions constrained by Dirichlet priors. BioMiCo is supervised, using known features for samples and appropriate prior constraints to overcome the challenges posed by many variables, sparse data, and large numbers of rare species. The model is trained on a portion of the data, where it learns how assemblages of species are mixed to form communities and how assemblages are related to the known features of each sample. Training yields a model that can predict the features of new samples. We used BioMiCo to build models for three serially sampled datasets and tested their predictive accuracy across different time points. The first model was trained to predict both body site (hand, mouth, and gut) and individual human host. It was able to reliably distinguish these features across different time points. The second was trained on vaginal microbiomes to predict both the Nugent score and individual human host. We found that women having normal and elevated Nugent scores had distinct microbiome structures that persisted over time, with additional structure within women having elevated scores. The third was trained for the purpose of assessing seasonal transitions in a coastal bacterial community. Application of this model to a high-resolution time series permitted us to track the rate and time of community succession and accurately predict known ecosystem-level events.ConclusionBioMiCo provides a framework for learning the structure of microbial communities and for making predictions based on microbial assemblages. By training on carefully chosen features (abiotic or biotic), BioMiCo can be used to understand and predict transitions between complex communities composed of hundreds of microbial species.Electronic supplementary materialThe online version of this article (doi:10.1186/s40168-015-0073-x) contains supplementary material, which is available to authorized users.

Characterising and predicting cyanobacterial blooms in an 8-year amplicon sequencing time course

“…To classify bloom and non-bloom samples (Supplementary Table 7), we used the Bayesian inference of microbial communities (BIOMICO) model described by Shafiei et al (2015). This supervised machine learning approach infers how OTUs are combined into assemblages and how combinations of these assemblages differ between bloom and non-bloom samples.…”

Latent Dirichlet Allocation reveals spatial and taxonomic structure in a DNA‐based census of soil biodiversity from a tropical forest

“…The popular software "Structure" of population genetics uses an almost identical model to characterize population structure based on the distribution of alleles across individuals (Pritchard, Stephens, & Donnelly, 2000). Several mixture models related to LDA have been used to investigate the diversity and structure of human microbiota based on the distribution of OTUs across individuals (Ding & Schloss, 2014;Holmes et al, 2012;Knights et al, 2011;Sankaran & Holmes, 2017;Shafiei et al, 2015). LDA has also been applied to ecology by Valle, Baiser, Woodall, and Chazdon (2014) and, in a slightly different version, by White, Dey, Mohan, Stephens, and Price (2019).…”

Dirichlet‐multinomial modelling outperforms alternatives for analysis of microbiome and other ecological count data

“…The success of DMM for relative abundance estimation, as demonstrated herein, coupled with the aforementioned benefits of hierarchical Bayesian modelling, justifies extension of the DMM to determine the effects of covariates on relative abundances and to characterize mixtures of compositions (sensu Chen & Li, 2013;Holmes et al, 2012;Knights et al, 2011;Shafiei et al, 2015;Tang & Chen, 2018). We look forward to continued method development along these lines.…”

“…Similar models have been applied to large counts of DNA sequences—for instance, Fernandes et al (; aldex2 ), Nowicka and Robinson (; drim‐seq ), and Rosa et al (; hmp ) use DMM to estimate and compare feature‐specific relative abundances in transcriptomes and microbiomes. Additionally, DMM has been used to model mixtures of compositions, a situation that could arise in a laboratory‐derived microbial assemblage occurring as a contaminant within samples, or in mixtures of different communities in nature ( microbedmm , Holmes, Harris, & Quince, ; sourcetracker , Knights et al, ; biomico , Shafiei et al, ; feast , Shenhav et al, ; ecostructure , White, Dey, Mohan, Stephens, & Price, ). Likewise, DMM has been used to estimate association networks among microbial taxa ( sparcc , Friedman & Alm, ; mldm , Yang, Chen, & Chen, ).…”