Multimodal surveillance of SARS-CoV-2 at a university enables development of a robust outbreak response framework.

BACKGROUND: Universities are vulnerable to infectious disease outbreaks, making them ideal environments to study transmission dynamics and evaluate mitigation and surveillance measures. Here, we analyze multimodal COVID-19-associated data collected during the 2020-2021 academic year at Colorado Mesa University and introduce a SARS-CoV-2 surveillance and response framework. METHODS: We analyzed epidemiological and sociobehavioral data (demographics, contact tracing, and WiFi-based co-location data) alongside pathogen surveillance data (wastewater and diagnostic testing, and viral genomic sequencing of wastewater and clinical specimens) to characterize outbreak dynamics and inform policy. We applied relative risk, multiple linear regression, and social network assortativity to identify attributes or behaviors associated with contracting SARS-CoV-2. To characterize SARS-CoV-2 transmission, we used viral sequencing, phylogenomic tools, and functional assays. FINDINGS: Athletes, particularly those on high-contact teams, had the highest risk of testing positive. On average, individuals who tested positive had more contacts and longer interaction durations than individuals who never tested positive. The distribution of contacts per individual was overdispersed, although not as overdispersed as the distribution of phylogenomic descendants. Corroboration via technical replicates was essential for identification of wastewater mutations. CONCLUSIONS: Based on our findings, we formulate a framework that combines tools into an integrated disease surveillance program that can be implemented in other congregate settings with limited resources. FUNDING: This work was supported by the National Science Foundation, the Hertz Foundation, the National Institutes of Health, the Centers for Disease Control and Prevention, the Massachusetts Consortium on Pathogen Readiness, the Howard Hughes Medical Institute, the Flu Lab, and the Audacious Project.

Detecting Intra-Field Variation in Rice Yield With Unmanned Aerial Vehicle Imagery and Deep Learning.

Unmanned aerial vehicles (UAVs) equipped with multispectral sensors offer high spatial and temporal resolution imagery for monitoring crop stress at early stages of development. Analysis of UAV-derived data with advanced machine learning models could improve real-time management in agricultural systems, but guidance for this integration is currently limited. Here we compare two deep learning-based strategies for early warning detection of crop stress, using multitemporal imagery throughout the growing season to predict field-scale yield in irrigated rice in eastern Arkansas. Both deep learning strategies showed improvements upon traditional statistical learning approaches including linear regression and gradient boosted decision trees. First, we explicitly accounted for variation across developmental stages using a 3D convolutional neural network (CNN) architecture that captures both spatial and temporal dimensions of UAV images from multiple time points throughout one growing season. 3D-CNNs achieved low prediction error on the test set, with a Root Mean Squared Error (RMSE) of 8.8% of the mean yield. For the second strategy, a 2D-CNN, we considered only spatial relationships among pixels for image features acquired during a single flyover. 2D-CNNs trained on images from a single day were most accurate when images were taken during booting stage or later, with RMSE ranging from 7.4 to 8.2% of the mean yield. A primary benefit of convolutional autoencoder-like models (based on analyses of prediction maps and feature importance) is the spatial denoising effect that corrects yield predictions for individual pixels based on the values of vegetation index and thermal features for nearby pixels. Our results highlight the promise of convolutional autoencoders for UAV-based yield prediction in rice.

Using Targeted Sequencing of Paralogous Sequences for Noninvasive Detection of Selected Fetal Aneuploidies.

BACKGROUND: Current methods for noninvasive prenatal testing (NIPT) ascertain fetal aneuploidies using either direct counting measures of DNA fragments from specific genomic regions or relative measures of single nucleotide polymorphism frequencies. Alternatively, the ratios of paralogous sequence pairs were predicted to reflect fetal aneuploidy. We developed a NIPT assay that uses paralog sequences to enable noninvasive detection of fetal trisomy 21 (T21) and trisomy 18 (T18) using cell-free DNA (cfDNA) from maternal plasma. METHODS: A total of 1060 primer pairs were designed to determine fetal aneuploidy status, fetal sex, and fetal fraction. Each library was prepared from cfDNA by coamplifying all 1060 target pairs together in a single reaction well. Products were measured using massively parallel sequencing and deviations from expected paralog ratios were determined based on the read depth from each paralog. RESULTS: We evaluated this assay in a blinded set of 480 cfDNA samples with fetal aneuploidy status determined by the MaterniT21® PLUS assay. Samples were sequenced (mean = 2.3 million reads) with 432 samples returning a result. Using the MaterniT21 PLUS assay for paired plasma aliquots from the same individuals as a reference, all 385 euploid samples, all 31 T21 samples, and 14 of 16 T18 samples were detected with no false positive results observed. CONCLUSIONS: This study introduces a novel NIPT aneuploidy detection approach using targeted sequencing of paralog motifs and establishes proof-of-concept for a potentially low-cost, highly scalable method for the identification of selected fetal aneuploidies with performance and nonreportable rate similar to other published methods.