Functional Biogeography of Ocean Microbes Revealed through Non-Negative Matrix Factorization

Abstract

The direct “metagenomic” sequencing of genomic material from complex assemblages of bacteria, archaea, viruses and microeukaryotes has yielded new insights into the structure of microbial communities. For example, analysis of metagenomic data has revealed the existence of previously unknown microbial taxa whose spatial distributions are limited by environmental conditions, ecological competition, and dispersal mechanisms. However, differences in genotypes that might lead biologists to designate two microbes as taxonomically distinct need not necessarily imply differences in ecological function. Hence, there is a growing need for large-scale analysis of the distribution of microbial function across habitats. Here, we present a framework for investigating the biogeography of microbial function by analyzing the distribution of protein families inferred from environmental sequence data across a global collection of sites. We map over 6,000,000 protein sequences from unassembled reads from the Global Ocean Survey dataset to protein families, generating a protein family relative abundance matrix that describes the distribution of each protein family across sites. We then use non-negative matrix factorization (NMF) to approximate these protein family profiles as linear combinations of a small number of ecological components. Each component has a characteristic functional profile and site profile. Our approach identifies common functional signatures within several of the components. We use our method as a filter to estimate functional distance between sites, and find that an NMF-filtered measure of functional distance is more strongly correlated with environmental distance than a comparable PCA-filtered measure. We also find that functional distance is more strongly correlated with environmental distance than with geographic distance, in agreement with prior studies. We identify similar protein functions in several components and suggest that functional co-occurrence across metagenomic samples could lead to future methods for de-novo functional prediction. We conclude by discussing how NMF, and other dimension reduction methods, can help enable a macroscopic functional description of marine ecosystems.

Microbial ecology of Antarctic aquatic systems

“…Overall, the study demonstrated that environmental temperature influenced the distribution of phylotypes, and hence the diversification and global radiation of the SAR11 clade. These findings corroborate other reports that identify temperature as a factor that controls phylotypes and the global distribution of marine microorganisms 29,32,35,[42][43][44][45][46] , including a study using single-cell genomics 47 .…”

Fine-scale spatial patterns in bacterial community composition and function within freshwater ponds

“…The relationships between microbial community composition, diversity and functional potential have been explored in several aquatic bacterial community studies (Langenheder et al, 2005;Szabo et al, 2007;Comte and del Giorgio, 2010;Peter et al, 2011). However, such studies are rarely undertaken in natural systems (but see Parnell et al, 2010;Jiang et al, 2012) and are often conducted as analyses of 'island-like' systems such as batch culture microcosms (Langenheder et al, 2005) where species richness is frequently lowered artificially (Peter et al, 2011;Ylla et al, 2013) or in poorly connected environments where the exchange of microbial taxa among microcosms is otherwise impeded. Thus, it remains unclear the extent to which microbial communities exhibit biogeography in taxonomic or functional attributes across small spatial scales within contiguous landscapes.…”

Inferring Aggregated Functional Traits from Metagenomic Data Using Constrained Non-negative Matrix Factorization: Application to Fiber Degradation in the Human Gut Microbiota

“…Parameter selection procedures proposed in most applications of NMF to biological data are based upon stability of the weight matrix W over several initializations of the iterative algorithm used to solve the problem [12, 13, 15, 16], in particular regarding the clustering of samples. Since our use of NMF focuses on extracting reproducible biological mechanisms characterized by the trait matrix H , we are more interested in biological stability than numerical one.…”

“…Since our use of NMF focuses on extracting reproducible biological mechanisms characterized by the trait matrix H , we are more interested in biological stability than numerical one. Therefore, we propose an adaptation of the concordance index developed by [16], which measures the concordance between two matrices of profiles, to evaluate the concordance of the CAFTs computed on independent data sets. In our approach, the reproducibility of H on a new data set is mimicked by repeatedly splitting the set of biological samples, performing the NMF decomposition on each subset and evaluating the similarity between the two trait matrices via the concordance index.…”

“…In particular Jiang and co-authors used this approach as a “soft” clustering tool in different ecosystems. They studied pathway abundances in diverse environmental ecosystems [15], compared habitats in marine ecosystems using protein families profiles [16] and human body sites using phylogenetic and functional data [17]. …”