UTUPub
Not a member yet
55238 research outputs found
Sort by
Dynamic adaptation of phytoplankton vertical migration to changing grazing and nutrient conditions
Get PDFDespite the ubiquitous occurrence of vertical migration of phytoplankton its quantitative significance remains poorly known. We eliminated a dense Daphnia population in a pond by introducing whitefish fingerlings, and assessed the effects on the vertical migration of dominating motile phytoplankton. At the highest abundance of Daphnia, cryptophytes reduced grazing losses by staying in the hypolimnion day and night, but Mallomonas species armoured by silica bristles remained in the epilimnion. After the fish introduction, phytoplankton was released from Daphnia grazing pressure, allowing cryptophytes to occur in the epilimnion also at noon. At the same time, increased phytoplankton biomass exacerbated the nutrient depletion. Cryptophytes compensated for that by migrating into the anoxic hypolimnion, whereupon their growth rates increased. The collapse of Daphnia was also associated with a temporary increase in nutrient regeneration by enzyme activities and decreases in total nutrient concentration and bacterial biomass in the whole water column. Our results show that cryptophytes can dynamically modify their vertical migration to balance between the exploitation of various nutrient resources and the risk of becoming eaten. Hypolimnetic nutrient resources can be quantitatively more important for phytoplankton than previously assumed.</p
Anthropogenic bottom-up and top-down impacts on boreal breeding waterbirds
Get PDFWetland habitats are changing under multiple anthropogenic pressures. Nutrient leakage and pollution modify physico-chemical state of wetlands and affect the ecosystem through bottom-up processes, while alien predators affect the ecosystems in a top-down manner. Boreal wetlands are important breeding areas for several waterbird species, the abundances of which potentially reflect both bottom-up and top-down ecosystem processes. Here, we use long-term national monitoring data gathered from c. 130 waterbird breeding sites in Finland from the 1980s to the 2020s. We hypothesised that the physico-chemical state of the waters and increasing alien predator abundance both play a role in steering the waterbird population trends. We set out to test this hypothesis by relating population changes of 17 waterbird species to changes in water chemistry and to regional alien predator indices while allowing species-specific effects to vary with foraging niche (dabblers, invertivore divers, piscivorous divers, herbivores), nesting site, female mass and habitat (oligotrophic, eutrophic). We found niche and nesting site-specific, habitat-dependent changes in waterbird numbers. While the associations with higher phosphorus levels and browning water were in overall positive at the oligotrophic lakes, the numbers of invertivore and piscivore diving ducks were most strongly negatively associated with higher phosphorus levels and browning water at the eutrophic lakes. Furthermore, increased pH levels benefitted piscivores. Invertivore diving duck species nesting on the wetlands had declined most on sites with high alien predator indices. Large herbivorous species and species preferring oligotrophic lakes seem to be successful. We conclude that the large-scale breeding waterbird decline in Finland is closely connected to both bottom-up and top-down processes, where negative associations are emphasised especially at eutrophic lakes. Niche-, nest site- and habitat-specific management actions are required to conserve declining waterbird populations. Managing wetlands on catchments level together with alien predator control may provide important approaches to future wetland management.</p
Brightest cluster galaxies: the centre can(not?) hold
Get PDFWe explore the persistence of the alignment of brightest cluster galaxies (BCGs) with their local environment. We find that a significant fraction of BCGs do not coincide with the centroid of the X-ray gas distribution and/or show peculiar velocities (they are not at rest with respect to the cluster mean). Despite this, we find that BCGs are generally aligned with the cluster mass distribution even when they have significant offsets from the X-ray centre and significant peculiar velocities. The large offsets are not consistent with simple theoretical models. To account for these observations BCGs must undergo mergers preferentially along their major axis, the main infall direction. Such BCGs may be oscillating within the cluster potential after having been displaced by mergers or collisions, or the dark matter halo itself may not yet be relaxed
The effect of neural network architecture on virtual H&E staining: Systematic assessment of histological feasibility
Get PDFConventional histopathology has relied on chemical staining for over a century. The staining process makes tissue sections visible to the human eye through a tedious and labor-intensive procedure that alters the tissue irreversibly, preventing repeated use of the sample. Deep learning-based virtual staining can potentially alleviate these shortcomings. Here, we used standard brightfield microscopy on unstained tissue sections and studied the impact of increased network capacity on the resulting virtually stained H&E images. Using the generative adversarial neural network model pix2pix as a baseline, we observed that replacing simple convolutions with dense convolution units increased the structural similarity score, peak signal-to-noise ratio, and nuclei reproduction accuracy. We also demonstrated highly accurate reproduction of histology, especially with increased network capacity, and demonstrated applicability to several tissues. We show that network architecture optimization can improve the image translation accuracy of virtual H&E staining, highlighting the potential of virtual staining in streamlining histopathological analysis.</p
Metabolic engineering of Streptomyces peucetius for biosynthesis of N,N-dimethylated anthracyclines
Get PDFIntroduction: Daunorubicin and doxorubicin, two anthracycline polyketides produced by S. peucetius, are potent anticancer agents that are widely used in chemotherapy, despite severe side effects. Recent advances have highlighted the potential of producing improved derivatives with reduced side effects by incorporating l-rhodosamine, the N,N-dimethyl analogue of the native amino sugar moiety.Method: In this study, we aimed to produce N,N-dimethylated anthracyclines by engineering the doxorubicin biosynthetic pathway in the industrial Streptomyces peucetius strain G001. To achieve this, we introduced genes from the aclarubicin biosynthetic pathway encoding the sugar N-methyltransferases AclP and AknX2. Furthermore, the native gene for glycosyltransferase DnrS was replaced with genes encoding the aclarubicin glycosyltransferases AknS and AknT. Additionally, the gene for methylesterase RdmC from the rhodomycin biosynthetic pathway was introduced.Results: A new host was engineered successfully, whereby genes from the aclarubicin pathway were introduced and expressed. LC-MS/MS analysis of the engineered strains showed that dimethylated sugars were efficiently produced, and that these were incorporated ino the anthracycline biosynthetic pathway to produce the novel dimethylated anthracycline N,N-dimethyldaunorubicin. Further downstream tailoring steps catalysed by the cytochrome P450 monooxygenase DoxA exhibited limited efficacy with N,N-dimethylated substrates. This resulted in only low production levels of N,N-dimethyldaunorubicin and no N,N-dimethyldoxorubicin, most likely due to the low affinity of DoxA for dimethylated substrates.Discussion: S. peucetius G001 was engineered such as to produce N,N-dimethylated sugars, which were incorporated into the biosynthetic pathway. This allowed the successful production of N,N-dimethyldaunorubicin, an anticancer drug with reduced cytotoxicity. DoxA is the key enzyme that determines the efficiency of the biosynthesis of N,N-dimethylated anthracyclines, and engineering of this enzyme will be a major step forwards towards the efficient production of more N,N-dimethylated anthracyclines, including N,N-dimethyldoxorubicin. This study provides valuable insights into the biosynthesis of clinically relevant daunorubicin derivatives, highlighting the importance of combinatorial biosynthesis.</p
Distributed Denial of Service Attack Detection for the Internet of Things Using Hybrid Deep Learning Model
Get PDFAs a result of the widespread adoption of the Internet of Things, there are now hundreds of millions of connected devices, increasing the likelihood that they may be vulnerable to various types of cyberattacks. In recent years, distributed denial of service (DDoS) has emerged as one of the most destructive tools utilized by attackers. Traditional machine learning approaches are typically ineffective and unable to cope with actual traffic properties when used to identify DDoS attacks. This paper introduces a novel deep learning-based intrusion detection system, specifically designed for deployment at either the Cloud or Fog level in the IoT environment. The proposed model aims to detect all types of DDoS attacks with their specific subcategory. Our hybrid model combines different types of deep learning models, including Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM), Deep Autoencoder, and Deep Neural Networks (DNNs). Our proposed model is made up of two main levels. The first one contains different parallel sub-neural networks trained with specific algorithms. The second level uses the output of the frozen first level combined with the initial data as input. The idea behind the combination of these various types of deep neural networks is to exploit their different properties to achieve very high performance. To evaluate our model, we used the CIC-DDoS2019 dataset, which satisfies all the constraints of an intrusion detection dataset. The results obtained demonstrate that our proposed model outperformed various well-known machine learning and deep learning models in terms of the true positive rate, accuracy, false alarm rate, average accuracy, and average detection rate.</p
Active learning of molecular data for task-specific objectives
Get PDFActive learning (AL) has shown promise to be a particularly data-efficient machine learning approach. Yet, its performance depends on the application, and it is not clear when AL practitioners can expect computational savings. Here, we carry out a systematic AL performance assessment for three diverse molecular datasets and two common scientific tasks: compiling compact, informative datasets and targeted molecular searches. We implemented AL with Gaussian processes (GP) and used the many-body tensor as molecular representation. For the first task, we tested different data acquisition strategies, batch sizes, and GP noise settings. AL was insensitive to the acquisition batch size, and we observed the best AL performance for the acquisition strategy that combines uncertainty reduction with clustering to promote diversity. However, for optimal GP noise settings, AL did not outperform the randomized selection of data points. Conversely, for targeted searches, AL outperformed random sampling and achieved data savings of up to 64%. Our analysis provides insight into this task-specific performance difference in terms of target distributions and data collection strategies. We established that the performance of AL depends on the relative distribution of the target molecules in comparison to the total dataset distribution, with the largest computational savings achieved when their overlap is minimal.</p
Contextuality with Disturbance and without: Neither Can Violate Substantive Requirements the Other Satisfies
Get PDFLipid and volatile profiles of Finnish oat batches of pure cultivars : Effect of storage on the volatile formation
Get PDFRecent data showing the compositional variation and storage behavior among different oat batches for the purpose of food remains limited. Lipids of twenty oat flour samples of pure cultivars grown in Finland during 2019 were extracted and fractionated into neutral and polar-rich lipids. Flour was stored for nine months, and profiles of volatiles and tocols were analyzed to reveal oxidative stability. The lipid content was 5.9–8.9 g per 100 g of flour [DW] and consisted of 78.7 ± 2.5 % neutral and 21.3 ± 2.5 % polar lipids. Palmitic (16 %), oleic (36 %), and linoleic (39 %) acids were the most abundant fatty acids. Neutral lipids had more oleic and less linoleic and palmitic acids than polar lipids. The fresh samples correlated with tocols, pentanal, 2-pentylfuran, 2-heptanone, nonanal, 2-butanone, and heptanal, while stored samples were associated with 3-octen-2-one, 2-octenal, hexanal, and octanal. Lipid composition and oxidative stability are essential factors for selecting oat batches for food applications