The expanding incorporation of bacterial whole-genome sequencing (WGS) into Public Health laboratories and regulatory agencies has enhanced foodborne outbreak detection and source attribution. By consequence, increasing volumes of publicly available whole-genome datasets can be directly used to investigate the population structure and epidemiology of foodborne pathogens (e.g., Salmonella enterica). Specifically, the mining of thousands of bacterial genomes over space and time can aid in harnessing population-based patterns, for which the frequency distribution of lineages/variants can be used as a trait. As such, we can use a hierarchical population structure analysis to infer genotypes at different levels of resolution, while selecting the most appropriate stratum to infer ecological signal and past selection. In order to optimize and facilitate scalable population genomic analysis of foodborne pathogens, we have recently developed the computational platform called ProkEvo. ProkEvo allows for scalable, reproducible, and automated population genomic analysis of pathogens, such as S. enterica, while facilitating pan-genomic mapping onto the hierarchical population structure. By combining the mining of share and sparse contents of the bacterial pan-genome, we can leverage the hierarchical population structure to 1) find the appropriate level of genotypic resolution to be focused on given the epidemiological context; 2) map the accessory-genome to infer Ecotypes and past selection in the population (or selectable units); and 3) identify genomic units that can be used to track variants of interest such as antimicrobial resistant (AMR) clones. In this talk, multiple case studies from S. enterica populations will be used to show how to assess population-based patterns in publicly available datasets and using a more systematically based sampling done across the food chain. In particular, it will show how three deterministic models of population diversification can be identified (convergent, divergent, and admixture), while inferring potential ecological traits that could have facilitated its formation. Ultimately, our group seeks leverage population genomic analysis to identify actionable knowledge that can be used or implemented by Public Health laboratories and regulatory agencies to improve food safety.

Weaning is a period of environmental changes and stress that results in significant alterations to the piglet microbiome and a predisposition to disease. We assessed the bacteriome and mycobiome of the gastrointestinal tract of piglets throughout the weaning transition to assess cross-kingdom interactions that alter animal health, growth, and disease-resistance. Feces and organs showed a dramatic shift over time in microbial communities, as well as an increase in network connectivity between kingdoms. The bacteriome showed a predictable pattern of development from Bacteroidaceae to Prevotellaceae, while the mycobiome demonstrated a loss in diversity with a Saccharomycetaceae-dominated post-weaning population. The mycobiome demonstrated a transient community that is driven by factors such as diet or environment rather than an organized pattern of colonization and succession. SparCC analyses found significant correlations between fungal and bacterial genera. These interactions were supported in vitro through bacterial isolates altering biofilm complexity of Kazachstania slooffiae, a porcine fungus. Correlation analyses linked specific taxa and plasma enteroendocrine peptides with piglet growth rate at specific ages, suggesting the potential for targeted probiotics/prebiotics. Ongoing work is utilizing diverse datasets from large animal cohort studies to develop machine learning algorithms that can better predict animal performance and molecules and/or microbes of interest.

Three important wine parameters: vineyard, region, and vintage year were evaluated using fifteen Vitis vinifera L. ‘Pinot noir’ wines derived from the same scion clone (Pinot noir 667). These wines were produced from two vintage years (2015 and 2016) and eight different regions along the Pacific Coast of the United States. We successfully improved the classification of the selected Pinot noir wines by combining untargeted 1D 1H NMR analysis with a targeted peptide based differential sensing array. NMR spectroscopy was used to evaluate the chemical fingerprint of the wines, while the peptide-based sensing array was utilized to mimic the senses of taste, smell, and palate texture by characterizing the phenolic profile. Multivariate multiblock analyses in the MVAPACK software package were utilized to combine the NMR and differential sensing array datasets. The regions of interest in both the NMR spectra and phenolic profiles in the peptide-based sensors were further analyzed through univariate analysis in the MetaboAnalyst software package. The combination of these chemometric analyses were utilized to improve the classification of genetically identical Pinot noir wines on the basis of distinctive signatures associated with the vineyard, region of growth, and vintage year.

Porcine reproductive and respiratory syndrome virus (PRRSV) causes substantial loss to the swine ind...

Phenotyping allows the measurement of morphometric and physiological parameters of plants in a rapid, non-destructive, accurate, and high-throughput manner. Traditional phenotyping in breeding is time-consuming, labor-intensive, and the database is insufficient to satisfy the needs of plant breeders which hampers the breeding progress. Recent advancements in electronics, and sensor technologies in agriculture have aided in the development of innovative methods for measuring phenotypic characteristics. These sensor systems can provide a high spatial and temporal resolution data to characterize crop growth parameters within the diverse environmental condition. In this study the Raspberry Pi (RPi) -based sensor imaging system was integrated with a camera (RPi Sony 8MP) in growth chamber to analyze the crop growth conditions in wheat breeding trial for automated phenotypic application. The collected digital images were suitable for extracting measureable plant traits. The plant traits studied includes morphometric parameters such as plant density, canopy cover, leaf area index, and physiological parameters such as photosynthetic rate and biomass, which represents the plant growth and health. The developed low cost digital imaging system will be integrated with internet to facilitate internet-of-things (IoT) based sensor system which helps plant breeder to make timely decisions, screen elite cultivar and monitor crop in real-time.

TBD

Crop diseases are a major threat to human food security. Around the world, more than 80% of agricultural production is generated by farmers, and over 50% of their yield is lost due to pests and pathogens, leading to mass disruption in food supply and a large number of hungry people. The purpose of this research was to create a free, easy-to-use, and widely accessible mobile application that efficiently and accurately, diagnoses 26 diseases of 14 crop species. Furthermore, this application provides treatment steps, common symptoms, and access to recommended curing products for each disease. The real-time crop disease diagnosis is based on a convolutional neural network (CNN) that was trained, validated, and tested on a dataset of 87,860 leaf images split into 38 classes. To design an optimal CNN, 16 different CNNs were designed and tested. MobileNetV2 using the Canny Edge Detection filter was chosen as it had the highest classification accuracy of 95.7 % and an Fl score of 96.1. This application is a novel and accessible tool for crop disease management and can be deployed as a free service to farmers for ecologically sustainable production, overall increasing food security.

Soil microbial communities mediate nutrient cycling, plant stress tolerance, and several other important ecological processes. Although DNA-based analyses have shed light upon the soil microbial community’s staggering diversity and functional potential, it can be difficult to distinguish active microbes from the background of dormant organisms and relic DNA present in soil. DNA stable isotope probing (SIP) can identify active organisms by tracing microbial incorporation of an isotopically enriched substrate into newly synthesized DNA. We employed quantitative SIP (qSIP) with 18O-enriched H2O to investigate bacterial response to water limitation in the presence and absence of two plant-associated fungal lineages (the arbuscular mycorrhizal fungus Rhizophagus irregularis and the Sebacinales fungus Serendipita bescii) grown with the bioenergy model grass Panicum hallii. In microcosms that were not inoculated with R. irregularis or S. bescii, a history of water limitation resulted in significantly lower bacterial growth rates, growth efficiency, and diversity within the actively growing bacterial community. In contrast, both fungi appeared to protect bacterial communities: bacterial growth rates, growth efficiency, and the diversity of the active bacterial community were not suppressed by a history of water limitation in soils inoculated with either fungus. Several of the bacterial taxa that responded positively to R. irregularis or S. bescii in water-limited soil belong to lineages that are considered susceptible to drought. qSIP highlighted effects of moisture history and fungal inoculum that were less pronounced in traditional 16S rRNA gene profile analysis. Together, these results suggest that plant-associated fungi support bacterial resilience to moisture limitation, and that qSIP may help reveal microbial interactions that support resilient agricultural systems.

The advent of fully automated data recording technologies and high-throughput phenotyping (HTP) systems has opened up a myriad of opportunities to advance breeding programs and livestock husbandry. Such technologies allow scoring large number of animals for novel phenotypes and indicator traits to boost genetic improvement. HTP tools comprise, for example, image analysis and computer vision, sensor technology for motion, sound and chemical composition, and spectroscopy. Potential applications of such technologies include recording individual feed intake for computing feed efficiency, social behavior and wellbeing, and various product quality traits. However, implementing HTP requires sophisticated statistical and computational approaches for efficient data management and appropriate data mining, as it involves large datasets with many covariates and complex relationships. In this talk we will discuss some of the challenges and potentials of HTP in animal breeding and genetics, and present examples of applications in beef and dairy cattle and pigs.

The purpose of the Functional Annotation of ANimal Genomes (FAANG) project is to catalog RNA-expressing regions and regulatory elements (REs) in agriculturally important animals. Such annotation is needed to understand the biological effect of genetic variants associated with economic traits for genetic improvement. Further, there is interest in integrating such functional data across biological states and across species to improve our understanding of comparative biology and genomics. In the domestic pig, we have collected RNA and epigenomics data on >20 adult and fetal tissues as well as nine circulating immune cell populations. We will focus today on an analysis of recent immune cell transcriptomic and epigenomics data (ATACseq, DNA methylation, histone modifications) collected on these cell populations to identify novel cell type-specific REs. Integration of deep epigenomics data using ChromHMM to identify chromatin states in cell populations identifies potential REs across the pig genome. We are also using single-cell (sc) RNAseq and scATACseq data on circulating immune cells to identify such REs. Interestingly, analysis of open chromatin regions identified by scATACseq identify similar and potentially more discriminative cell types than does scRNAseq data. Integration of these data through correlation of RE activity with cell-type specific promoter element activity can predict putative regulatory networks controlling cell type-specific expression. Thus, these data are the foundation needed to create useful regulatory networks that map functional elements to the target genes they regulate. This regulatory information will add to the value of FAANG data for understanding the function of the genome and application to genetic improvement of domesticated animals.

Targeted engineering of plant genomes holds great promise for ensuring food security and for producing biopharmaceuticals. However, this engineering requires thorough knowledge of cis-regulatory elements to precisely control endogenous and introduced genes. To generate this knowledge, we established plant STARR-seq, a massively parallel reporter assay, which can measure the condition-specific activity of hundreds of thousands of putative regulatory elements.
We used plant STARR-seq to characterize over 75,000 promoters from Arabidopsis, maize and sorghum. We demonstrate that core promoter elements as well as GC content and transcription factor binding sites influence promoter strength. By performing the experiments in two assay systems, leaves of the dicot tobacco and protoplasts of the monocot maize, we detect species-specific differences. Using these observations, we built computational models to predict promoter strength in both assay systems, allowing us to design highly active synthetic promoters comparable in activity to the viral 35S minimal promoter.
We recently assessed enhancer activity for over 175,000 accessible chromatin regions from Arabidopsis, tomato, maize and sorghum, in addition to testing almost 1,000,000 random sequences. We show that enhancers are orientation-independent and that their strength is determined by transcription factor binding sites and GC content. By testing these elements in different environmental conditions, we identify both constitutive and condition-specific enhancers and determine the features that are responsible for their activity. Using this data, we have trained computational models to accurately predict and design condition-specific enhancers. Notably, our previous promoter models have little power to predict enhancer activity, and promoter strength alone was not predictive of endogenous gene expression. We are currently combining the knowledge we have gained for both element types to derive models that can predict endogenous gene expression, identify promising targets for genome engineering, and predict the outcome of genomic edits. Together with novel plant terminators and insulators – whose activity we are presently measuring – this comprehensive strategy will enable us to build tunable and programmable multi-gene cassettes that encode metabolic pathways to produce valuable bioproducts in crops.

With rapid reductions in sequencing costs and increased accessibility to high throughput phenotyping, large multi-dimensional datasets are now available for specialty crops such as strawberry. The strawberry breeding program at the University of Florida has generated population-level datasets for SNPs, short-read resequencing, RNA-seq, and remote-sensing phenotypes. These datasets have provoked new research questions and provided new insights into the genetic architecture and prediction of agronomic and fruit quality traits. Several impacts and future prospects will be highlighted. In one application, we integrated phased genome assemblies, an eQTL map, and a structural variant map to resolve candidate genes and their regulatory elements for flavor compounds. In another example, Bayesian and machine learning models facilitated genomic selection (GS) with high accuracy in multiple years for independent test populations. These GS approaches are now routinely applied in parental selection to reduce the breeding cycle by one year, as well as within seedling crosses using low marker-densities to increase selection intensity. Finally, field imagery and spectral reflectance data were combined with SNP data to predict plant biomass and powdery mildew resistance, demonstrating the power of a combined genomic/phenomic selection approach in breeding.

In its current form, field phenotyping with a focus on genotype-to-phenotype analysis is both laborious and limited. One major limitation is in how the data are collected. We generally use specific protocols for measuring and documenting plant phenotypes, and where the phenotypes are descriptive, controlled vocabularies and ontologies are used to constrain the descriptions to a consistent and computable format. We aim to expand researchers’ “toolbox” for field-based phenotyping by developing methods to collect spoken descriptive phenotypes using natural language. In brief, we transcribe speech to text using pre-existing platforms, then process the data to derive semantic similarity. From there, we generate networks and identify highly interconnected clusters, which we call synthetic phenotypes, then use those clusters along with genotype data for association mapping. As a proof of concept, we designed an experiment that compares spoken phenotype descriptions to traditional trait measurement by student researchers using the Wisconsin Diversity panel (grown in Boone, Iowa summer 2021). In this seminar, I will describe how we approached each component of the system and review the current status and next steps.