Dataset Information

Machine learning to predict microbial community functions: An analysis of dissolved organic carbon from litter decomposition.

ABSTRACT: Microbial communities are ubiquitous and often influence macroscopic properties of the ecosystems they inhabit. However, deciphering the functional relationship between specific microbes and ecosystem properties is an ongoing challenge owing to the complexity of the communities. This challenge can be addressed, in part, by integrating the advances in DNA sequencing technology with computational approaches like machine learning. Although machine learning techniques have been applied to microbiome data, use of these techniques remains rare, and user-friendly platforms to implement such techniques are not widely available. We developed a tool that implements neural network and random forest models to perform regression and feature selection tasks on microbiome data. In this study, we applied the tool to analyze soil microbiome (16S rRNA gene profiles) and dissolved organic carbon (DOC) data from a 44-day plant litter decomposition experiment. The microbiome data includes 1709 total bacterial operational taxonomic units (OTU) from 300+ microcosms. Regression analysis of predicted and actual DOC for a held-out test set of 51 samples yield Pearson's correlation coefficients of.636 and.676 for neural network and random forest approaches, respectively. Important taxa identified by the machine learning techniques are compared to results from a standard tool (indicator species analysis) widely used by microbial ecologists. Of 1709 bacterial taxa, indicator species analysis identified 285 taxa as significant determinants of DOC concentration. Of the top 285 ranked features determined by machine learning methods, a subset of 86 taxa are common to all feature selection techniques. Using this subset of features, prediction results for random permutations of the data set are at least equally accurate compared to predictions determined using the entire feature set. Our results suggest that integration of multiple methods can aid identification of a robust subset of taxa within complex communities that may drive specific functional outcomes of interest.

SUBMITTER: Thompson J

PROVIDER: S-EPMC6602172 | biostudies-literature | 2019

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine learning to predict microbial community functions: An analysis of dissolved organic carbon from litter decomposition.

Thompson Jaron J Johansen Renee R Dunbar John J Munsky Brian B

PloS one 20190701 7

Microbial communities are ubiquitous and often influence macroscopic properties of the ecosystems they inhabit. However, deciphering the functional relationship between specific microbes and ecosystem properties is an ongoing challenge owing to the complexity of the communities. This challenge can be addressed, in part, by integrating the advances in DNA sequencing technology with computational approaches like machine learning. Although machine learning techniques have been applied to microbiome ...[more]

PMID: 31260460

Similar Datasets

Project description:Knowledge of dynamic interactions between natural organic matter (NOM) and microbial communities is critical not only to delineate the routes of NOM degradation/transformation and carbon (C) fluxes, but also to understand microbial community evolution and succession in ecosystems. Yet, these processes in subsurface environments are usually studied independently, and a comprehensive view has been elusive thus far. In this study, we fed sediment-derived dissolved organic matter (DOM) to groundwater microbes and continually analyzed microbial transformation of DOM over a 50-day incubation. To document fine-scale changes in DOM chemistry, we applied high-resolution Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) and soft X-ray absorption spectroscopy (sXAS). We also monitored the trajectory of microbial biomass, community structure and activity over this time period. Together, these analyses provided an unprecedented comprehensive view of interactions between sediment-derived DOM and indigenous subsurface groundwater microbes. Microbial decomposition of labile C in DOM was immediately evident from biomass increase and total organic carbon (TOC) decrease. The change of microbial composition was closely related to DOM turnover: microbial community in early stages of incubation was influenced by relatively labile tannin- and protein-like compounds; while in later stages the community composition evolved to be most correlated with less labile lipid- and lignin-like compounds. These changes in microbial community structure and function, coupled with the contribution of microbial products to DOM pool affected the further transformation of DOM, culminating in stark changes to DOM composition over time. Our study demonstrates a distinct response of microbial communities to biotransformation of DOM, which improves our understanding of coupled interactions between sediment-derived DOM, microbial processes, and community structure in subsurface groundwater.

Project description:Mixing plant litters of multiple species can alter litter decomposition, a key driver of carbon and nutrient cycling in terrestrial ecosystems. Changes in microbial decomposer communities is proposed as one of the mechanisms explaining this litter-mixture effect, but the underlying mechanism is unclear. In a microcosm litterbag experiment, we found that, at the early stage of decomposition, litter mixing promoted tomato root litter decomposition, thus generating a synergistic nonadditive litter-mixture effect. The transplanting decomposer community experiment showed that changes in microbial decomposer communities contributed to the nonadditive litter-mixture effect on tomato root litter decomposition. Moreover, litter mixing altered the abundance and diversity of bacterial and fungal communities on tomato root litter. Litter mixing also stimulated several putative keystone operational taxonomic units (OTUs) in the microbial correlation network, such as Fusarium sp. fOTU761 and Microbacterium sp. bOTU6632. Then, we isolated and cultured representative isolates of these two taxa, named Fusarium sp. F13 and Microbacterium sp. B26. Subsequent in vitro tests found that F13, but not B26, had strong decomposing ability; moreover, these two isolates developed synergistic interaction, thus promoted litter decomposition in coculture. Addition of F13 or B26 both promoted the decomposing activity of the resident decomposer community on tomato root litter, confirming their importance for litter decomposition. Overall, litter mixing promoted tomato root litter decomposition through altering microbial decomposers, especially through stimulating certain putative keystone taxa. IMPORTANCE Microbial decomposer community plays a key role in litter decomposition, which is an important regulator of soil carbon and nutrient cycling. Though changes in decomposer communities has been proposed as one of the potential underlying mechanisms driving the litter-mixture effects, direct evidence is still lacking. Here, we demonstrated that litter mixing stimulated litter decomposition through altering microbial decomposers at the early stage of decomposition. Moreover, certain putative keystone taxa stimulated by litter mixing contributed to the nonadditive litter-mixture effect. In vitro culturing validated the role of these taxa in litter decomposition. This study also highlights the possibility of regulating litter decomposition through manipulating certain microbial taxa.

Project description:What is the effect of phyllosphere microorganisms on litter decomposition in the absence of colonization by soil microorganisms? Here, we simulated the litter standing decomposition stage in the field to study the differences in the composition and structure of the phyllosphere microbial community after the mixed decomposition of Populus × canadensis and Pinus sylvestris var. mongolica litter. After 15 months of mixed decomposition, we discovered that litters that were not in contact with soil had an antagonistic effect (the actual decomposition rate was 18.18%, which is lower than the expected decomposition rate) and the difference between the litters themselves resulted in a negative response to litter decomposition. In addition, there was no significant difference in bacterial and fungal community diversity after litter decomposition. The litter bacterial community was negatively responsive to litter properties and positively responsive to the fungal community. Importantly, we found that bacterial communities had a greater impact on litter decomposition than fungi. This study has enriched our understanding of the decomposition of litter itself and provided a theoretical basis for further exploring the "additive and non-additive effects" of litter decomposition and the mechanism of microbial drive.ImportanceThe study of litter decomposition mechanism plays an important role in the material circulation of the global ecosystem. However, previous studies have often looked at contact with soil as the starting point for decomposition. But actually, standing litter is very common in forest ecosystems. Therefore, we used field simulation experiments to simulate the decomposition of litters without contact with soil for 15 months, to explore the combined and non-added benefits of the decomposition of mixed litters, and to study the influence of microbial community composition on the decomposition rate while comparing the differences of microbial communities.

Dataset Information

Machine learning to predict microbial community functions: An analysis of dissolved organic carbon from litter decomposition.

Publications

Machine learning to predict microbial community functions: An analysis of dissolved organic carbon from litter decomposition.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets