Bioinformatics Platform of Plant Fat Related Genes, Proteins and Metabolism

Increasing seed oil content for biofuels and bioproducts through breeding and biotechnology often leads to trade-offs, such as reduced protein content, smaller seed size, or lower seed fitness. The molecular basis of these trade-offs remains unclear. Efforts to understand the metabolic effects of altered oil content and composition, as well as yield reductions, are often complicated by off-target genetic mutations, making cause-and-effect relationships difficult to interpret. To address these challenges, we propose a comprehensive, integrated strategy to study the effects of increased and customized lipid production in transgenic plants specifically engineered to alter seed oil content and composition. As an extension of a prior project, we are developing FatPlants (www.fatplants.net), a community-driven web platform that serves as a centralized resource for data related to modifying oil composition and enhancing oil content in plants. This platform integrates curated public datasets from funded websites and literature with new data generated from this project. As part of a collaborative effort, FatPlants also incorporates data from two other plant lipid-focused websites: ARALIP and PMN. The platform currently supports multiple crop and model oilseed species, including Camelina sativa (camelina), Thlaspi arvense (pennycress), and Cuphea viscosissima, an "extreme" producer of medium-chain fatty acid-rich seeds. FatPlants provides a comprehensive repository of all known fatty acid-related genes and proteins in these species and overlays these data with lipidomic measurements from B5 target species. To facilitate comparative analysis, FatPlants offers a suite of bioinformatics tools, including a pathway viewer, protein structure viewer, BLAST search, protein-protein interaction viewer, and GO enrichment analysis. Additionally, to enhance collaboration among B5 investigators, the platform includes a secure, user-authenticated internal data-sharing space for participating research labs. FatPlants is publicly available as a community resource at www.fatplants.net

Nitrogen dioxide (NO₂) is a harmful air pollutant that can cause various health issues. Monitoring of NO₂ mostly dependent on expensive ground-based sensor systems. This research explored the potential of integrating Sentinel-2 and Sentinel-5P to estimate high resolution ground NO₂ concentration at city and neighborhood level. This study presented convolutional neural networks (CNN) based a two-stream deep learning model architecture for NO₂ estimation. The experimental results show that model performs better if it utilizes data from both sources. While model performance was assessed on datasets aggregated in three temporal frequencies, the estimated NO₂ was found more reliable for quarterly level followed by monthly. The daily estimates are more variable due to short-term fluctuation of NO₂ concentration. This study reinforces the application of deep learning and remote sensing for air quality monitoring especially in absence of expensive ground-based monitoring systems.