Next generation sequencing (NGS) combined with high-resolution bioinformatics, offers powerful method for detection, identification, and characterization of pathogenic microorganisms (bacteria, viruses, fungi, and parasites). This approach to diagnosis of infectious disease agents and infectious diseases offers accuracy, speed, and actionable information, the sequencing within one or two days and bioinformatics analysis within minutes. We have applied this method in clinical studies, including retrospective case control studies comprising samples known and unknown etiology, as well samples from healthy individuals. The results are exciting and demonstrate detection and identification of pathogens can be accomplished well within the time frame of a single day or so. Furthermore, microbiome analysis can be used to differentiate healthy, diseased, and asymptomatic carriers, including individuals in early stages of infection and disease. Results of studies accomplished to date show that disease state of patients reveals multiple pathogens, the microbial communities of healthy humans of diverse geographic locations tolerate different levels of pathogenic microorganisms and antibiotic resistance in their microbiomes. Different rates of antibiotic resistance were detected in geographically diverse populations. Analysis of the human microbiome and the microbial ecology of aquatic reservoirs and wastewater reuse treatment plants has provided insight into the complex interactions of the microbial populations of these ecosystems. Precision offered by next generation sequencing coupled with powerful bioinformatics makes possible a much more complete understanding of the microbiology of human populations and their environment. Thus, it is clear that a Genomics Center at North South University is both timely and valuable resource for Bangladesh.

The opportunities for distilling microbiome data into commercial products are surfacing rapidly. However, data distilleries are faced with challenges in broad variations in biospecimen preparation, assay selection, bioinformatics pipelines and statistical approaches, as evidenced by discordant microbiome observations found in the literature. Awareness of these sources of variation enables well-structured machine learning at cloud compute scale to find diagnostic biomarkers of disease, beneficial microbes and therapeutic microbial secretions. Furthermore, the benefits of adding PhyloChip analysis on top of NGS workflows will be discussed, with special attention to the benefits of zero barcodes, zero multiplexing, internal controls within each sample, automation, low technical variation and production of non-sparse matrices. Integration of NGS and PhyloChip data into a unified bioinformatics analysis allows confident product candidate nominations.

This session will discuss the microbiome as a potential lens through which to view the trajectory of a complex disease. Using Parkinson’s disease as an example, we will discuss the environmental factors that can potentially impact the microbiome and the course of disease. This session will discuss The BioCollective’s unique approach to collecting broad-based population samples from healthy and disease specific groups with the goal of collecting, connecting and correcting the microbiome to change the course of chronic disease. Dr. Yates will present data on collection and preservation methods to maximize culturing.

Soil microorganisms interact with root surfaces (rhizosphere) and within root tissues (endosphere), affecting host plant productivity and driving soil biogeochemistry. Zea mays L. (corn) and Glycine max L. (soybean) together are the most dominant agricultural products globally. The root-associated microbiomes of these crops can therefore influence microbial biodiversity and functional capacity of soil on a global scale. However, in-depth characterization of these microbiomes currently does not exist. In this study, we sampled corn and soybean root associated microbiomes (rhizosphere and endosphere) across central USA through amplicon sequencing (16S rRNA) and shotgun metagenomics. Using a novel iterative assembly approach based on the bacterial diversity of each sample, we generated high quality reference metagenomes for corn and soy root-associated microbiomes that are the largest to date. Soy microbiomes were enriched in Rhizobiales and nitrogen fixation genes following classical models of legume-rhizobia mutualisms. The root-associated microbiomes of corn were more diverse and enriched in a variety of nutrient acquisition strategies, suggesting that multiple resource limitation creates more niche space for microorganisms to coexist. Our reference metagenomes represent a publicly available resource that will assist in determining microbial-based targets for improving agricultural productivity and sustainability.

Culture-independent sequencing (metagenomics) is a powerful, high resolution technique enabling the study of microbial communities in situ. With modern sequencing technology and bioinformatics, individual genomes can be assembled and extracted directly from environmental samples containing complex microbial communities by a process known as metagenomic "binning." However, available binning programs suffer from methodological and practical shortcomings, such as the requirement of human pattern recognition, which is inherently unscalable, low-throughput, and poorly reproducible. Some methods also require the assembly of pooled samples, which can lead to poor assemblies in the case of inter-sample strain variability. We therefore devised a fully-automated pipeline, termed "Autometa," which incorporates machine learning principles to separate pure microbial genomes from single shotgun metagenomes. Autometa uses Barnes-Hut Stochastic Neighbor Embedding to analyze 5-mer frequency in the contiguous sequences (i.e., "contigs") produced by de novo metagenomic assembly. The DBSCAN algorithm is then used to identify groups of contigs (i.e., genome "bins") with congruent 5-mer frequency patterns. Unsupervised machine learning is then employed to optimize clustering for purity of genome bins, measured by the presence of gene markers known to occur as single copies in isolated strains. In preliminary tests, Autometa recovered more pure and complete genomes from simulated, synthetic, and environmental metagenomic samples as compared to available programs such as MaxBin and MetaBAT. We are actively integrating supervised machine learning to further refine the binning process and using our current implementation of Autometa to study natural product biosynthesis in marine invertebrate microbiomes.