Abteilung Umweltchemie

Convolutional Neural Networks (CNNs) offer an alternative to the image cross-correlation methods used in Particle Image Velocimetry (PIV) to reconstruct the fluid velocity field from the experimental recording. Despite the flexibility of CNNs, the accuracy and robustness of the standard image processing remains unsurpassed for general PIV data. As CNNs are non-linear and typically entail up to millions of trainable parameters, they require large and carefully designed training datasets to avoid over-fitting and to obtain results that are accurate for a wide range of flow conditions and length scales. Most training datasets consist of PIV-like data that are generated from displacement fields resulting from numerical flow simulations, which, in addition of being computationally expensive, may be able to inform the network only about relatively few classes of flow problems. To overcome this issue and improve the accuracy of the velocity reconstructed by CNNs, we propose to train the networks with synthetic PIV-like data generated from random displacement fields. The underlying idea is that the training dataset simply needs to teach the network about the kinematic relationship between position and velocity. These kinematic training datasets are computationally inexpensive and may allow a much richer variability in terms of length scales by varying the generation parameters. By training a state-of-the-art CNN, we investigate the accuracy of the reconstructed displacement and velocity with synthetic and experimental test cases, such as a sinusoidal flow and wind-tunnel data from a turbulent-boundary-layer and a cylinder-wake experiment. We demonstrate that kinematic training can drastically improve the accuracy of the CNN and allows the network to outperform conventional cross-correlation methods, being more robust with respect to data noise and providing reconstructed velocity fields that have considerably higher spatial resolution (at pixel level).

Over the past three decades, the Vietnamese Mekong Delta has experienced a significant increase in agricultural productivity, partly achieved through increased agrochemical use. To abate negative effects on human and environmental health, several national programs were launched to enhance safer pesticide use. This study aimed to assess the patterns and relationships of official sustainable agriculture educational programs, pesticide safety knowledge, and practices of smallholder farmers in the Mekong Delta. A cross-sectional survey was conducted with 400 smallholder farmers from three communes in Thoi Lai district (Can Tho province) from March to May 2020. Twenty-four questions on pesticide safety knowledge and practices were used to identify traits using latent class analysis. Adjusted generalized linear regression was used to assess determinants of pesticide safety knowledge and estimate associations of pesticide safety knowledge with pesticide practices. 96.2% of participants have used at least one WHO class II pesticide during the past year while the use of specific personal protective equipment was limited mainly due to unavailability (37.0%) or discomfort (83.0%). High education (Odds Ratio (OR), 95% Confidence Interval; 3.84, 1.70-9.45), exposure to official educational programs (1.87, 1.13-3.12), peer-to-peer knowledge exchange (3.58, 2.18-6.00), and learning from governmental extension services (2.31, 1.14-4.98) were positively associated with increased pesticide safety knowledge. Compared to poor practices, pesticide safety knowledge was increasingly positively associated with intermediate (1.65, 1.02-2.66) and good pesticide practices (8.96, 2.58-31.12). These findings highlight the importance of school education and educational programs, access to PPE, and addressing discomforts of PPE to improve the protection of farmers from pesticide exposures. Simultaneously, pesticide market authorization processes should be reconsidered to promote the authorization of less toxic products. Further in-depth studies on the nature of pesticides used, nonuse of personal protective equipment, and effectiveness of educational programs will further define leverage points for safer pesticide use.

Oxygenations of aromatic soil and water contaminants with molecular O₂ catalyzed by Rieske dioxygenases are frequent initial steps of biodegradation in natural and engineered environments. Many of these non-heme ferrous iron enzymes are known to be involved in contaminant metabolism, but the understanding of enzyme-substrate interactions that lead to successful biodegradation is still elusive. Here, we studied the mechanisms of O₂ activation and substrate hydroxylation of two nitroarene dioxygenases to evaluate enzyme- and substrate-specific factors that determine the efficiency of oxygenated product formation. Experiments in enzyme assays of 2-nitrotoluene dioxygenase (2NTDO) and nitrobenzene dioxygenase (NBDO) with methyl-, fluoro-, chloro-, and hydroxy-substituted nitroaromatic substrates reveal that typically 20-100% of the enzyme's activity involves unproductive paths of O₂ activation with generation of reactive oxygen species through so-called O₂ uncoupling. The ¹⁸O and ¹³C kinetic isotope effects of O₂ activation and nitroaromatic substrate hydroxylation, respectively, suggest that O₂ uncoupling occurs after generation of Fe^III-(hydro)peroxo species in the catalytic cycle. While 2NTDO hydroxylates ortho-substituted nitroaromatic substrates more efficiently, NBDO favors meta-substituted, presumably due to distinct active site residues of the two enzymes. Our data implies, however, that the O₂ uncoupling and hydroxylation activity cannot be assessed from simple structure-reactivity relationships. By quantifying O₂ uncoupling by Rieske dioxygenases, our work provides a mechanistic link between contaminant biodegradation, the generation of reactive oxygen species, and possible adaptation strategies of microorganisms to the exposure of new contaminants.

Current methods for structure elucidation of small molecules rely on finding similarity with spectra of known compounds, but do not predict structures de novo for unknown compound classes. We present MSNovelist, which combines fingerprint prediction with an encoder–decoder neural network to generate structures de novo solely from tandem mass spectrometry (MS²) spectra. In an evaluation with 3,863 MS² spectra from the Global Natural Product Social Molecular Networking site, MSNovelist predicted 25% of structures correctly on first rank, retrieved 45% of structures overall and reproduced 61% of correct database annotations, without having ever seen the structure in the training phase. Similarly, for the CASMI 2016 challenge, MSNovelist correctly predicted 26% and retrieved 57% of structures, recovering 64% of correct database annotations. Finally, we illustrate the application of MSNovelist in a bryophyte MS² dataset, in which de novo structure prediction substantially outscored the best database candidate for seven spectra. MSNovelist is ideally suited to complement library-based annotation in the case of poorly represented analyte classes and novel compounds.