Machine Learning Predicts Biogeochemistry from Microbial Community Structure in a Complex Model System.

Microbial community structure is influenced by the environment and in turn exerts control on many environmental parameters. We applied this concept in a bioreactor study to test whether microbial community structure contains information sufficient to predict the concentration of H2S as the product of sulfate reduction. Microbial sulfate reduction is a major source of H2S in many industrial and environmental systems and is often influenced by the existing physicochemical conditions. Production of H2S in industrial systems leads to occupational hazards and adversely affects the quality of products. A long-term (148 days) experiment was conducted in upflow bioreactors to mimic sulfidogenesis, followed by inhibition with nitrate salts and a resumption of H2S generation when inhibition was released. We determined microbial community structure in 731 samples across 20 bioreactors using 16S rRNA gene sequencing and applied a random forest algorithm to successfully predict different phases of sulfidogenesis and mitigation (accuracy = 93.17%) and sessile and effluent microbial communities (accuracy = 100%). Similarly derived regression models that also included cell abundances were able to predict H2S concentration with remarkably high fidelity (R2 > 0.82). Metabolic profiles based on microbial community structure were also found to be reliable predictors for H2S concentration (R2 = 0.78). These results suggest that microbial community structure contains information sufficient to predict sulfidogenesis in a closed system, with anticipated applications to microbially driven processes in open environments. IMPORTANCE Microbial communities control many biogeochemical processes. Many of these processes are impractical or expensive to measure directly. Because the taxonomic structure of the microbial community is indicative of its function, it encodes information that can be used to predict biogeochemistry. Here, we demonstrate how a machine learning technique can be used to predict sulfidogenesis, a key biogeochemical process in a model system. A distinction of this research was the ability to predict H2S production in a bioreactor from the effluent bacterial community structure without direct observations of the sessile community or other environmental conditions. This study establishes the ability to use machine learning approaches in predicting sulfide concentrations in a closed system, which can be further developed as a valuable tool for predicting biogeochemical processes in open environments. As machine learning algorithms continue to improve, we anticipate increased applications of microbial community structure to predict key environmental and industrial processes.

Detection of Sulfate-Reducing Bacteria as an Indicator for Successful Mitigation of Sulfide Production.

Sulfate-reducing bacteria (SRBs) are one of the main sources of biogenic H2S generation in oil reservoirs. Excess H2S production in these systems leads to oil biosouring, which causes operational risks and health hazards and can increase the cost of refining crude oil. Nitrate salts are often added to the system to suppress sulfidogenesis. Because SRB populations can persist in biofilms even after nitrate treatment, identifying shifts in the sessile community is crucial for successful mitigation. However, sampling the sessile community is hampered by its inaccessibility. Here, we use the results of a long-term (148 days) ex situ experiment to identify particular sessile community members from observations of the sample waste stream. Microbial community structure was determined for 731 samples across 20 bioreactors using 16S rRNA gene sequencing. By associating microbial community structure with specific steps in the mitigation process, we could distinguish between taxa associated with H2S production and mitigation. After initiation of nitrate treatment, certain SRB populations increased in the planktonic community during critical time points, indicating the dissociation of SRBs from the biofilm. Predicted relative abundances of the dissimilatory sulfate reduction pathway also increased during the critical time points. Here, by analyzing the planktonic community structure, we describe a general method that uses high-throughput amplicon sequencing, metabolic inferences, and cell abundance data to identify successful biofilm mitigation. We anticipate that our approach is also applicable to other systems where biofilms must be mitigated but cannot be sampled easily. IMPORTANCE Microbial biofilms are commonly present in many industrial processes and can negatively impact performance and safety. Within the oil industry, subterranean biofilms cause biosouring with implications for oil quality, cost, occupational health, and the environment. Because these biofilms cannot be sampled directly, methods are needed to indirectly assess the success of mitigation measures. This study demonstrates how the planktonic microbial community can be used to assess the dissociation of sulfate-reducing bacterium (SRB)-containing biofilms. We found that an increase in the abundance of a specific SRB population in the effluent after nitrate treatment can be used as a potential indicator for the successful mitigation of biofilm-forming SRBs. Moreover, a method for determining critical time points for detecting potential indicators is suggested. This study expands our knowledge of improving mitigation strategies for biosouring and could have broader implications in other systems where biofilms lead to adverse consequences.

Understanding Microbial Community Dynamics in Up-Flow Bioreactors to Improve Mitigation Strategies for Oil Souring.

Oil souring occurs when H2S is generated in oil reservoirs. This not only leads to operational risks and health hazards but also increases the cost of refining crude oil. Sulfate-reducing microorganisms are considered to be the main source of the H2S that leads to oil souring. Substrate competition between nitrate-reducing and sulfate-reducing microorganisms makes biosouring mitigation via the addition of nitrate salts a viable strategy. This study explores the shift in microbial community across different phases of biosouring and mitigation. Anaerobic sand-filled columns wetted with seawater and/or oil were used to initiate the processes of sulfidogenesis, followed by mitigation with nitrate, rebound sulfidogenesis, and rebound control phases (via nitrate and low salinity treatment). Shifts in microbial community structure and function were observed across different phases of seawater and oil setups. Marine bacterial taxa (Marinobacter, Marinobacterium, Thalassolituus, Alteromonas, and Cycloclasticus) were found to be the initial responders to the application of nitrate during mitigation of sulfidogenesis in both seawater- and oil- wetted columns. Autotrophic groups (Sulfurimonas and Desulfatibacillum) were found to be higher in seawater-wetted columns compared to oil-wetted columns, suggesting the potential for autotrophic volatile fatty acid (VFA) production in oil-field aquifers when seawater is introduced. Results indicate that fermentative (such as Bacteroidetes) and oil-degrading bacteria (such as Desulfobacula toluolica) play an important role in generating electron donors in the system, which may sustain biosouring and nitrate reduction. Persistence of certain microorganisms (Desulfobacula) across different phases was observed, which may be due to a shift in metabolic lifestyle of the microorganisms across phases, or zonation based on nutrient availability in the columns. Overall results suggest mitigation strategies for biosouring can be improved by monitoring VFA concentrations and microbial community dynamics in the oil reservoirs during secondary recovery of oil.

Combined genomarkers approach to Salmonella characterization reveals that nucleotide sequence differences in the phase 1 flagellin gene fliC are markers for variation within serotypes.

The characterization and tracking of pathogenic micro-organisms in the clinical laboratory and public health environment demand schemes that are easy to standardize and use, are automated and high-throughput, and provide portable data. A combined genomarkers approach for Salmonella enterica based on comparative sequence analysis by mass spectrometry has been developed. The scheme targets genes encoding synthesis and assembly of antigens, metabolic pathway enzymes, virulence factors and fluoroquinolone resistance, covering the essential sequences that distinguish between and identify variation within serotypes. This study demonstrated how this single method could replace the combination of methods currently required to determine serotypes, subtypes, antibiotic resistance profiles and the genomic relatedness of Salmonella isolates. The results revealed genomic variation within seven serotypes previously unreported. This variation can be detected by using nucleotide sequence differences in the Salmonella flagellin gene fliC as markers that are not detected by traditional serotyping methods.

Molecular complexity of successive bacterial epidemics deconvoluted by comparative pathogenomics.

Understanding the fine-structure molecular architecture of bacterial epidemics has been a long-sought goal of infectious disease research. We used short-read-length DNA sequencing coupled with mass spectroscopy analysis of SNPs to study the molecular pathogenomics of three successive epidemics of invasive infections involving 344 serotype M3 group A Streptococcus in Ontario, Canada. Sequencing the genome of 95 strains from the three epidemics, coupled with analysis of 280 biallelic SNPs in all 344 strains, revealed an unexpectedly complex population structure composed of a dynamic mixture of distinct clonally related complexes. We discovered that each epidemic is dominated by micro- and macrobursts of multiple emergent clones, some with distinct strain genotype-patient phenotype relationships. On average, strains were differentiated from one another by only 49 SNPs and 11 insertion-deletion events (indels) in the core genome. Ten percent of SNPs are strain specific; that is, each strain has a unique genome sequence. We identified nonrandom temporal-spatial patterns of strain distribution within and between the epidemic peaks. The extensive full-genome data permitted us to identify genes with significantly increased rates of nonsynonymous (amino acid-altering) nucleotide polymorphisms, thereby providing clues about selective forces operative in the host. Comparative expression microarray analysis revealed that closely related strains differentiated by seemingly modest genetic changes can have significantly divergent transcriptomes. We conclude that enhanced understanding of bacterial epidemics requires a deep-sequencing, geographically centric, comparative pathogenomics strategy.

Genomic and functional studies of Drosophila sex hierarchy regulated gene expression in adult head and nervous system tissues.

The Drosophila sex determination hierarchy controls all aspects of somatic sexual differentiation, including sex-specific differences in adult morphology and behavior. To gain insight into the molecular-genetic specification of reproductive behaviors and physiology, we identified genes expressed in the adult head and central nervous system that are regulated downstream of sex-specific transcription factors encoded by doublesex (dsx) and fruitless (fru). We used a microarray approach and identified 54 genes regulated downstream of dsx. Furthermore, based on these expression studies we identified new modes of DSX-regulated gene expression. We also identified 90 and 26 genes regulated in the adult head and central nervous system tissues, respectively, downstream of the sex-specific transcription factors encoded by fru. In addition, we present molecular-genetic analyses of two genes identified in our studies, calphotin (cpn) and defective proboscis extension response (dpr), and begin to describe their functional roles in male behaviors. We show that dpr and dpr-expressing cells are required for the proper timing of male courtship behaviors.