Jared Streich

Genetic mapping is a powerful tool for eukaryotic genetics that has only been applied to bacteria in limited circumstances. Quantitative trait locus (QTL) mapping generally relies on sexual recombination to break linkages between genes, yet bacteria rarely undergo sufficient homologous recombination to generate suitable mapping populations. In this work, we used iterative biparental genome shuffling by protoplast fusion in Bacillus subtilis to generate a population of bacteria with substantial random recombination throughout their genomes. Individual shuffled progeny were arrayed in well plates, resequenced, and characterized for a range of complex phenotypes, including spore germination and swarming motility. Genetic mapping of the resulting phenotypes identified high-confidence QTLs of moderate size (~15 kb), and these associations were validated through targeted genetic swaps. This B. subtilis QTL population can easily be used to map additional phenotypes and the general approach for QTL mapping is applicable to a wide range of Gram-positive and Gram-negative bacteria using diverse methods for genome-wide recombination.

Genetic mapping is a powerful tool for eukaryotic genetics that has only been applied to bacteria in limited circumstances. Quantitative trait locus (QTL) mapping generally relies on sexual recombination to break linkages between genes, yet bacteria rarely undergo sufficient homologous recombination to generate suitable mapping populations. In this work, we used iterative biparental genome shuffling by protoplast fusion in Bacillus subtilis to generate a population of bacteria with substantial random recombination throughout their genomes. Individual shuffled progeny were arrayed in well plates, resequenced, and characterized for a range of complex phenotypes including spore germination and swarming motility. Genetic mapping of the resulting phenotypes identified high-confidence QTLs of moderate size (∼10 kb), and these associations were validated through targeted genetic swaps. This B. subtilis QTL population can easily be used to map additional phenotypes, and the general approach for QTL mapping is applicable in a wide range of bacteria. One-Sentence Summary: Bacterial genome shuffling mimics sexual recombination and enables genetic mapping of complex phenotypes to causal alleles.

Plant phenotyping is typically a time-consuming and expensive endeavor, requiring large groups of researchers to meticulously measure biologically relevant plant traits, and is the main bottleneck in understanding plant adaptation and the genetic architecture underlying complex traits at population scale. In this work, we address these challenges by leveraging few-shot learning with convolutional neural networks to segment the leaf body and visible venation of 2,906 Populus trichocarpa leaf images obtained in the field. In contrast to previous methods, our approach (a) does not require experimental or image preprocessing, (b) uses the raw RGB images at full resolution, and (c) requires very few samples for training (e.g., just 8 images for vein segmentation). Traits relating to leaf morphology and vein topology are extracted from the resulting segmentations using traditional open-source image-processing tools, validated using real-world physical measurements, and used to conduct a genome-wide association study to identify genes controlling the traits. In this way, the current work is designed to provide the plant phenotyping community with (a) methods for fast and accurate image-based feature extraction that require minimal training data and (b) a new population-scale dataset, including 68 different leaf phenotypes, for domain scientists and machine learning researchers. All of the few-shot learning code, data, and results are made publicly available.

We present an exascale approach for producing global scale, high resolution, longitudinally based geoclimate classifications. Using a GPU implementation of the DUO Similarity Metric on the Summit supercomputer, we calculated the pairwise environmental similarity of 156,384,190 vectors of 414,640 encoded elements derived from 71 environmental variables over a 50-year time span at 1km ² resolution. GPU matrix-matrix (GEMM) kernels were optimized for the GPU architecture and their outputs were managed through aggressive concurrent MPI rank CPU communication, calculations, and transfers. Using vector transformation and highly optimized operations of generalized distributed dense linear algebra, calculation of all-vector-pairs similarity resulted in 5.07 × 10 ²¹ element comparisons and reached a peak performance of 2.31 exaflops. We demonstrated this method using existing and synthesized climate layers to show how geography can be parsed using high-performance computing. Geoclimate zones are important tools for understanding how environmental variables impact natural systems, particularly for agriculture and conservation with relevance to climate change. Historically, classification systems have been low resolution, based on limited variables, or subjective. To identify climate classes, we clustered DUO outputs at varying stringency, producing 69, 133, 340, and 717 global geoclimate zones. Our approach produced global scale, high resolution, longitudinally informed climate classifications that can be used in precision agriculture, cultivar breeding efforts, and conservation programs.

In this work we identify changes in high-resolution zones across the globe linked by environmental similarity that have implications for agriculture, bioenergy, and zoonosis. We refine exhaustive vector comparison methods with improved similarity metrics as well as provide multiple methods of amalgamation across 744 months of climatic data. The results of the vector comparison are captured as networks which are analyzed using static and longitudinal comparison methods to reveal locations around the globe experiencing dramatic changes in abiotic stress. Specifically we (i) incorporate updated similarity scores and provide a comparison between similarity metrics, (ii) implement a new feature for resource optimization, (iii) compare an agglomerative view to a longitudinal view, (iv) compare across 2-way and 3-way vector comparisons, (v) implement a new form of analysis, and (vi) demonstrate biological applications and discuss implications across a diverse set of species distributions by detecting changes that affect their habitats. Species of interest are related to agriculture (e.g., coffee, wine, chocolate), bioenergy (e.g., poplar, switchgrass, pennycress), as well as those living in zones of concern for zoonotic spillover that may lead to pandemics (e.g., eucalyptus, flying foxes).

Predicted growth in world population will put unparalleled stress on the need for sustainable energy and global food production, as well as increase the likelihood of future pandemics. In this work, we identify high-resolution environmental zones in the context of a changing climate and predict longitudinal processes relevant to these challenges. We do this using exhaustive vector comparison methods that measure the climatic similarity between all locations on Earth at high geospatial resolution relative to global-scale analyses. The results are captured as networks, in which edges between geolocations are defined if their historical climate similarities exceed a threshold. We apply Markov clustering and our novel correlation of correlations method to the resulting climatic networks, which provides unprecedented agglomerative and longitudinal views of climatic relationships across the globe. The methods performed here resulted in the fastest (9.37 × 10¹⁸ operations/s) and one of the largest 168.7 × 10²¹ operations) scientific computations ever performed, with more than 100 quadrillion edges considered for a single climatic network. Our climatic analysis reveals areas of the world experiencing rapid environmental changes, which can have important implications for global carbon fluxes and zoonotic spillover events. Correlation and network analyses of this kind are widely applicable across computational and predictive biology domains, including systems biology, ecology, carbon cycles, biogeochemistry, and zoonosis research.

Plant phenotyping is typically a time-consuming and expensive endeavor, requiring large groups of researchers to meticulously measure biologically relevant plant traits, and is the main bottleneck in understanding plant adaptation and the genetic architecture underlying complex traits at population scale. In this work, we address these challenges by leveraging few-shot learning with convolutional neural networks (CNNs) to segment the leaf body and visible venation of 2,906 P. trichocarpa leaf images obtained in the field. In contrast to previous methods, our approach (i) does not require experimental or image pre-processing, (ii) uses the raw RGB images at full resolution, and (iii) requires very few samples for training (e.g., just eight images for vein segmentation). Traits relating to leaf morphology and vein topology are extracted from the resulting segmentations using traditional open-source image-processing tools, validated using real-world physical measurements, and used to conduct a genome-wide association study to identify genes controlling the traits. In this way, the current work is designed to provide the plant phenotyping community with (i) methods for fast and accurate image-based feature extraction that require minimal training data, and (ii) a new population-scale data set, including 68 different leaf phenotypes, for domain scientists and machine learning researchers. All of the few-shot learning code, data, and results are made publicly available.

Despite a global decrease in suicide rates in recent years, death by suicide has increased in the United States. It is therefore imperative to identify the risk factors associated with suicide attempts in order to combat this growing epidemic. In this study, we use an explainable-artificial intelligence method, iterative Random Forest, to predict suicide attempts using data from the Million Veteran Program. Our predictive model incorporates multiple environmental variables (e.g., elevation, light wavelength absorbance, temperature, humidity, etc) at ZIP code-level geospatial resolution. We additionally consider demographic variables from the American Community Survey as well as the number of firearms and alcohol vendors per 10,000 people in order to assess the contributions of proximal environment, access to means, and restraint decrease to suicide attempts. Our results show that geographic areas with higher concentrations of married males living with spouses are predictive of lower rates of suicide attempts, whereas geographic areas where males are more likely to live alone and to rent housing are predictive of higher rates of suicide attempts. We also identified climatic features that were associated with suicide attempt risk by age group. Additionally, we observed that firearms and alcohol vendors were associated with increased risk for suicide attempts irrespective of the age group examined, but that their effects were small in comparison to the top features. Taken together, our findings highlight the importance of social determinants and environmental factors in understanding suicide risk among veterans.

In eukaryotes, fine-scale maps of meiotic recombination events have greatly advanced our understanding of the factors that affect genomic variation patterns and evolution of traits. However, in bacteria that lack natural systems for sexual reproduction, unbiased characterization of recombination landscapes has remained challenging due to variable rates of genetic exchange and influence of natural selection. Here, to overcome these limitations and to gain a genome-wide view on recombination, we crossed Bacillus strains with different genetic distances using protoplast fusion. The offspring displayed complex inheritance patterns with one of the parents consistently contributing the major part of the chromosome backbone and multiple unselected fragments originating from the second parent. Our results demonstrate that this bias was in part due to the action of restriction–modification systems, whereas genome features like GC content and local nucleotide identity did not affect distribution of recombination events around the chromosome. Furthermore, we found that recombination occurred uniformly across the genome without concentration into hotspots. Notably, our results show that species-level genetic distance did not affect genome-wide recombination. This study provides a new insight into the dynamics of recombination in bacteria and a platform for studying recombination patterns in diverse bacterial species.

Predicted growth in world population will put unparalleled stress on the need for sustainable energy and global food production, as well as increase the likelihood of future pandemics. In this work, we identify high-resolution environmental zones in the context of a changing climate and predict longitudinal processes relevant to these challenges. We do this using exhaustive vector comparison methods that measure the climatic similarity between all locations on earth at high geospatial resolution. The results are captured as networks, in which edges between geolocations are defined if their historical climates exceed a similarity threshold. We then apply Markov clustering and our novel Correlation of Correlations method to the resulting climatic networks, which provides unprecedented agglomerative and longitudinal views of climatic relationships across the globe. The methods performed here resulted in the fastest (9.37 × 10 ¹⁸ operations/sec) and one of the largest (168.7 × 10 ²¹ operations) scientific computations ever performed, with more than 100 quadrillion edges considered for a single climatic network. Correlation and network analysis methods of this kind are widely applicable across computational and predictive biology domains, including systems biology, ecology, carbon cycles, biogeochemistry, and zoonosis research.