Search

Identification of significant biological relationships or patterns is central to many metagenomic studies. Methods that estimate association networks have been proposed for this purpose; however, they assume that associations are static, neglecting the fact that relationships in a microbial ecosystem may vary with changes in environmental factors, which can result in inaccurate estimations. Therefore, this paper proposes a computational model, called the k-Lognormal-Dirichlet-Multinomial (kLDM) model, which estimates multiple association networks that correspond to specific environmental conditions, and simultaneously infers microbe-microbe and environmental factor-microbe associations for each network. The effectiveness of the kLDM model was demonstrated on synthetic data, a colorectal cancer (CRC) dataset, the Tara Oceans dataset, and the American Gut Project dataset. The results reveal that the widely-used Spearman's rank correlation coefficient (SCC) method performed much worse than the other methods, indicating the importance of separating samples by environmental conditions. Cancer fecal samples were then compared with cancer-free samples, and the estimation achieved by kLDM exhibited fewer associations among microbes but stronger associations between specific bacteria, especially five CRC-associated operational taxonomic units (OTUs), indicating gut microbe translocation in cancer patients. Some environmental factor-dependent associations were then found within a marine eukaryotic community. Finally, the gut microbial heterogeneity of irritable bowel disease (IBD) patients was detected. These results demonstrate that kLDM can elucidate the complex associations within microbial ecosystems. The kLDM program, R, and Python scripts, together with all experimental datasets, are accessible at https://github.com/tinglab/kLDM.git.

Human lifestyle and physiological variables on human disease risk have been revealed to be mediated by gut microbiota. Low concordance between many case-control studies for detecting disease-associated microbe existed and it is likely due to the limited sample size and the population-wide bias in human lifestyle and physiological variables. To infer association between whole gut microbiota and diseases accurately, we propose to build machine learning models by including both human variables and gut microbiota based on the American Gut Project data, the largest known publicly available human gut bacterial microbiota dataset. When the model's performance with both gut microbiota and human variables is better than the model with just human variables, the independent association of gut microbiota with the disease will be confirmed. We found that gut microbes showed different association strengths with different diseases. Adding gut microbiota into human variables enhanced the association strengths with inflammatory bowel disease (IBD) and unhealthy status; showed no effect on association strengths with Diabetes and IBS; reduced the association strengths with small intestinal bacterial overgrowth, C. difficile infection, lactose intolerance, cardiovascular disease and mental disorders. Our results suggested that although gut microbiota was reported to be associated with many diseases, a considerable proportion of these associations may be spurious. We also proposed a list of microbes as biomarkers to classify IBD and unhealthy status, and validated them by reference to previously published research.IMPORTANCEwe reexamined the association between gut microbiota and multiple diseases via machine learning models on a large-scale dataset, and by considering the effect of human variables ignored by previous studies, truly independent microbiota-disease associations were estimated. We found gut microbiota is associated independently with IBD and overall health of human, but more evidence is needed to judge associations between microbiota and other diseases. Further functional investigations of our reported disease-related microbes will improve understanding of the molecular mechanism of human diseases.

BackgroundFreshwater lakes are threatened by harmful cyanobacterial blooms; whose basic unit is Cyanobacterial Aggregate (CA). Community variations of CA-attached bacteria are substantial during different blooming stages. However, little is known about their transcriptional and metabolic variations. Most bacterial genomes in CA were not constructed in existing database, which limits our understanding of the bacterial variations as responses to cyanobacterial blooms. ResultsIn this longitudinal study, 16 CA samples were collected from Lake Taihu, one of the largest freshwater lakes in China, from April of 2015 to February of 2016. By sequencing the V4 region of 16S rRNA genes, full metagenomes (MG) and metatranscriptomes (MT), we generated 424 Mb of 16S rRNA gene data, 122 Gb of high-quality MG data and 160 Gb of high-quality MT data. We analyzed the taxonomic, functional and transcriptional variations of microbes in CAs along three blooming stages, and constructed metagenome-assembled genomes (MAGs) by binning analysis. First, 55 OTUs, 456 genes and 37 transcripts mainly associated with pathways of transporters, photosystem and energy metabolism showed significantly different abundance among the three stages. Second, 161 high-quality MAGs in CAs were achieved, with 19 of which significantly shifted in relative abundance among three stages. The most abundant MAGs have gene capacities to synthesize flagella and divers of transporters, and participate in metabolic pathways of nitrogen, phosphorus and sulfur. Finally, 22 high-quality cyanobacterial MAGs were constructed and can be divided into four functional clusters, which showed significant differences on the energy pathways, transporters and prokaryotic defense system.ConclusionOverall, these results demonstrated the taxonomic, functional and transcriptional variations of microbes in CAs among three different blooming stages. Genome construction and metabolic analysis of cyanobacteria and their attached bacteria suggested that the material exchange and signal transmission do, indeed, exist among them. Our understanding of the underlying molecular pathways for cyanobacterial blooms could potentially lead to the control of blooms by interventional strategies to disrupt the expression of critical microbes.

Identifying significant biological relationships or patterns is central to many metagenomic studies. Methods that estimate association networks have been proposed for this purpose, but they assume that associations are static, neglecting the fact that relationships in a microbial ecosystem may vary with changes in environmental factors, which can result in inaccurate estimations. We propose a computational model, k-Lognormal-Dirichlet-Multinomial model (kLDM), which estimates multiple association networks that correspond to specific environmental conditions according to values of environmental factors (EFs), and simultaneously infers microbe-microbe and environmental factor-microbe associations for each network. We showed the effectiveness of kLDM on synthetic data, a colorectal cancer dataset, as well as the TARA Oceans and American Gut project datasets. The results showed that the widely used Spearman’s rank correlation coefficient (SCC) performed much worse than other methods, indicating the importance of separating samples by environmental conditions. We compared cancer fecal samples with cancer-free samples, and our estimation showed fewer associations among microbes but stronger associations between specific bacteria such as five colorectal cancer (CRC)-associated OTUs, indicating gut microbe translocation in cancer patients. Some environmental-factor-dependent associations were found within marine eukaryotic community, and gut microbial heterogeneity of irritable bowel disease (IBD) patients was detected. Results demonstrated that kLDM could successfully unravel the underlying biological associations. In summary, our study presents a computational framework that can elucidate the complex associations within microbial ecosystems. The kLDM program, R, and python scripts, together with all experimental datasets are all accessible at Github (https://github.com/tinglab/kLDM.git).

Various types of diet can lead to type 2 diabetes. The gut microbiota in type 2 diabetic patients are also different. So, two questions arise: whether there are any commonalities between gut microbiota induced by different pro-obese diets and whether these commonalities lead to disease. Here we found that high-energy diets with two different fat-to-sugar ratios can both cause obesity and prediabetes but enrich different gut microbiota. Still, these different gut microbiota have similar genetic and metabolite compositions. The microbial metabolites in common between the diets modulate lipid accumulation and macrophage inflammation in vivo and in vitro. This work suggests that studies that only use 16S rRNA amplicon sequencing to determine how the microbes respond to diet and associate with diabetic state are missing vital information., Gut microbiota play important roles in host metabolism, especially in diabetes. However, why different diets lead to similar diabetic states despite being associated with different microbiota is not clear. Mice were fed two high-energy diets (HED) with the same energy density but different fat-to-sugar ratios to determine the associations between the microbiota and early-stage metabolic syndrome. The two diets resulted in different microbiota but similar diabetic states. Interestingly, the microbial gene profiles were not significantly different, and many common metabolites were identified, including l-aspartic acid, cholestan-3-ol (5β, 3α), and campesterol, which have been associated with lipogenesis and inflammation. Our study suggests that different metabolic-syndrome-inducing diets may result in different microbiota but similar microbiomes and metabolomes. This suggests that the metagenome and metabolome are crucial for the prognosis and pathogenesis of obesity and metabolic syndrome. IMPORTANCE Various types of diet can lead to type 2 diabetes. The gut microbiota in type 2 diabetic patients are also different. So, two questions arise: whether there are any commonalities between gut microbiota induced by different pro-obese diets and whether these commonalities lead to disease. Here we found that high-energy diets with two different fat-to-sugar ratios can both cause obesity and prediabetes but enrich different gut microbiota. Still, these different gut microbiota have similar genetic and metabolite compositions. The microbial metabolites in common between the diets modulate lipid accumulation and macrophage inflammation in vivo and in vitro. This work suggests that studies that only use 16S rRNA amplicon sequencing to determine how the microbes respond to diet and associate with diabetic state are missing vital information.

Freshwater lakes are threatened by harmful blooms characterized by Cyanobacterial Aggregates (CAs) that are normally aggregated with extracellular polysaccharides released by cyanobacteria to form a phycosphere. It is possible that mutualistic relationships exist between bacteria and cyanobacteria in these CAs wherein bacterial products supplement cyanobacterial growth, and cyanobacterial exudates, in turn, serve as substrates for bacteria, thus augmenting the stability of each constituent. However, little is known about the exact interaction between cyanobacteria and their attached bacteria in CAs. Therefore, in this study, we collected 26 CA samples from Lake Taihu, a large freshwater lake in China from March of 2015 to February of 2016. We then sequenced both the V4 regions of 16S rRNA genes and full metagenomes, resulting in 610 Mb of 16S rRNA gene data and 198.98 Gb of high-quality metagenomic data. We observed that two cyanobacteria genera (Microcystis and Dolichospermum) alternately dominated CAs along the sampling time and specific bacterial genera attached to different cyanobacteria genera dominated CAs. More specifically, Dolichospermum dominates CAs when water temperature is low and total nitrogen is high, while Microcystis dominates CAs when water temperature is high and total nitrogen is low. Moreover, we found specific bacterial genera attached to different cyanobacteria genera dominated CAs. The cyanobacteria-bacteria related pairs Dolichospermum-Burkholderia and Microcystis-Hyphomonas were detected by ecological networks construction. Bacterial communities in CAs were found to be more correlated with the cyanobacterial community (Mantel's r = 0.76, P = 0.001) than with environmental factors (Mantel's r = 0.27, P = 0.017). A potential codependent nitrogen-cycling pathway between cyanobacteria and their attached bacteria was constructed, indicating their functional link. Overall, these results demonstrated that mutualistic relationships do, indeed, exist between cyanobacteria and bacteria in CAs at both taxonomic and gene levels, providing biological clues potentially leading to the control of blooms by interventional strategies to disrupt bacteria-cyanobacteria relationships and co-pathways.

In aquatic ecosystems, both phytoplankton and bacteria play pivotal roles. Based on 16S rRNA gene sequencing, considerable research focused on phytoplankton colony attached and free-living bacteria has revealed the close relationship between them, and indicated that the entire bacterial community mediates crucial biogeochemical processes in aquatic ecosystems. However, our understanding of their distribution patterns and response to environmental factors remains poor. Besides, picocyanobacteria, which were omitted from attached bacteria analysis, were reported to be important in cyanobacterial blooms. To explore the spatiotemporal variation of the entire bacterial community with their driving environmental factors and detect the relationships among them, we collected 61 water samples spanning one year and the entire Lake Taihu regions for surveying the entire bacterial community. Our results indicated: 1) seasonal variation of the bacterial community composition was stronger than spatial variation due to the clearly seasonal variation of Microcystis, Synechococcus (pico-cyanobacteria) and other bacteria (Actinomycetales, Pirellulaceae and Sphingobacteriaceae); 2) the spatial distribution of the bacterial community showed that different phyla were dominant in different regions; 3) the bacterial co-occurrence networks varied seasonally and were dominated by Microcystis, ACK-M1, Chthoniobacteraceae, Synechococcus, Pirellulaceae and Pelagibacteraceae; 4) phytoplankton density, chlorophyll a, water temperature and total nitrogen were the major factors that drove the spatiotemporal variation of bacterial community composition. This study revealed the seasonal succession and spatial distribution of the entire bacterial community in Lake Taihu, providing new insights into the relationship between water bloom-forming cyanobacterial species and other bacteria, and their response to environmental factors in eutrophic freshwater ecosystem.

Massive partial sequencing of 16S rRNA genes has become the predominant tool used for studying microbial ecology. However, determining which hypervariable regions and primer sets should be used for screening microbial communities requires extensive investigation if controversial results are to be avoided. Here, the performances of different variable regions of the 16S rRNA gene on bacterial diversity studies were evaluated in silico with respect to the SILVA non-redundant reference database (SILVA SSU Ref 123NR), and subsequently verified using samples from Lake Taihu in China, a eutrophic lake. We found that the bacterial community composition results were strongly impacted by the different V regions. The results show that V1-V2 and V1-V3 regions were the most reliable regions in the full-length 16S rRNA sequences, while most V3 to V6 regions (including V3, V4, V3-V4, V5, V4-V5, V6, V3-V6, V4-V6, and V5-V6) were more closely aligned with the SILVA SSU Ref 123NR database. Overall, V4 was the most prominent V region for achieving good domain specificity, higher coverage and a broader spectrum in the Bacteria domain, as confirmed by the validation experiments. S-D-Bact-0564-a-S-15/S-D-Bact-0785-b-A-18 is, therefore, a promising primer set for surveying bacterial diversity in eutrophic lakes.