Atlantic cod individual spatial behaviour and stable isotope associations in a no-take marine reserve.

Foraging is a behavioural process and, therefore, individual behaviour and diet are theorized to covary. However, few comparisons of individual behaviour type and diet exist in the wild. We tested whether behaviour type and diet covary in a protected population of Atlantic cod, Gadus morhua. Working in a no-take marine reserve, we could collect data on natural behavioural variation and diet choice with minimal anthropogenic disturbance. We inferred behaviour using acoustic telemetry and diet from stable isotope compositions (expressed as δ13 C and δ15 N values). We further investigated whether behaviour and diet could have survival costs. We found cod with shorter diel vertical migration distances fed at higher trophic levels. Cod δ13 C and δ15 N values scaled positively with body size. Neither behaviour nor diet predicted survival, indicating phenotypic diversity is maintained without survival costs for cod in a protected ecosystem. The links between diet and diel vertical migration highlight that future work is needed to understand whether the shifts in this behaviour during environmental change (e.g. fishing or climate), could lead to trophic cascades.

Surface material of acoustic transmitters influences the inflammatory response of rainbow trout (Oncorhynchus mykiss) during long-term implantation.

Telemetry tags are a widely used technology for tracking animals that are difficult to observe in their natural environment. This technology has been increasingly used to monitor and study populations of high value salmonid species in Canadian waters. This study expands on a previous study of the impacts of tag implantation on the immune system of Rainbow Trout (Oncorhynchus mykiss). Pro-inflammatory cytokines and protein level markers were examined in fish that underwent peritoneal implantation of three tag types and compared to a sham surgery control group. The different materials on the surface of the tags showed differential immune induction extending over a two-month period. This included peritoneal total protein, IL-1ß protein, the immunoglobulins IgT and IgM, as well as pro-inflammatory transcripts in the spleen. These results are suggestive of a prolonged, costly foreign body response which may be differentially induced by the different types of tag coating, with ceramic tags being least immunogenic. Examining tag impacts at the level of the immune system will facilitate the development of more biocompatible tags which will improve data fidelity. This will support more effective strategies for the management of fisheries resources.

Evaluating the utility of brightfield image data for mechanism of action prediction.

Fluorescence staining techniques, such as Cell Painting, together with fluorescence microscopy have proven invaluable for visualizing and quantifying the effects that drugs and other perturbations have on cultured cells. However, fluorescence microscopy is expensive, time-consuming, labor-intensive, and the stains applied can be cytotoxic, interfering with the activity under study. The simplest form of microscopy, brightfield microscopy, lacks these downsides, but the images produced have low contrast and the cellular compartments are difficult to discern. Nevertheless, by harnessing deep learning, these brightfield images may still be sufficient for various predictive purposes. In this study, we compared the predictive performance of models trained on fluorescence images to those trained on brightfield images for predicting the mechanism of action (MoA) of different drugs. We also extracted CellProfiler features from the fluorescence images and used them to benchmark the performance. Overall, we found comparable and largely correlated predictive performance for the two imaging modalities. This is promising for future studies of MoAs in time-lapse experiments for which using fluorescence images is problematic. Explorations based on explainable AI techniques also provided valuable insights regarding compounds that were better predicted by one modality over the other.

A method for Boolean analysis of protein interactions at a molecular level.

Determining the levels of protein-protein interactions is essential for the analysis of signaling within the cell, characterization of mutation effects, protein function and activation in health and disease, among others. Herein, we describe MolBoolean - a method to detect interactions between endogenous proteins in various subcellular compartments, utilizing antibody-DNA conjugates for identification and signal amplification. In contrast to proximity ligation assays, MolBoolean simultaneously indicates the relative abundances of protein A and B not interacting with each other, as well as the pool of A and B proteins that are proximal enough to be considered an AB complex. MolBoolean is applicable both in fixed cells and tissue sections. The specific and quantifiable data that the method generates provide opportunities for both diagnostic use and medical research.

Predicting protein network topology clusters from chemical structure using deep learning.

Comparing chemical structures to infer protein targets and functions is a common approach, but basing comparisons on chemical similarity alone can be misleading. Here we present a methodology for predicting target protein clusters using deep neural networks. The model is trained on clusters of compounds based on similarities calculated from combined compound-protein and protein-protein interaction data using a network topology approach. We compare several deep learning architectures including both convolutional and recurrent neural networks. The best performing method, the recurrent neural network architecture MolPMoFiT, achieved an F1 score approaching 0.9 on a held-out test set of 8907 compounds. In addition, in-depth analysis on a set of eleven well-studied chemical compounds with known functions showed that predictions were justifiable for all but one of the chemicals. Four of the compounds, similar in their molecular structure but with dissimilarities in their function, revealed advantages of our method compared to using chemical similarity.

Learning to see colours: Biologically relevant virtual staining for adipocyte cell images.

Fluorescence microscopy, which visualizes cellular components with fluorescent stains, is an invaluable method in image cytometry. From these images various cellular features can be extracted. Together these features form phenotypes that can be used to determine effective drug therapies, such as those based on nanomedicines. Unfortunately, fluorescence microscopy is time-consuming, expensive, labour intensive, and toxic to the cells. Bright-field images lack these downsides but also lack the clear contrast of the cellular components and hence are difficult to use for downstream analysis. Generating the fluorescence images directly from bright-field images using virtual staining (also known as "label-free prediction" and "in-silico labeling") can get the best of both worlds, but can be very challenging to do for poorly visible cellular structures in the bright-field images. To tackle this problem deep learning models were explored to learn the mapping between bright-field and fluorescence images for adipocyte cell images. The models were tailored for each imaging channel, paying particular attention to the various challenges in each case, and those with the highest fidelity in extracted cell-level features were selected. The solutions included utilizing privileged information for the nuclear channel, and using image gradient information and adversarial training for the lipids channel. The former resulted in better morphological and count features and the latter resulted in more faithfully captured defects in the lipids, which are key features required for downstream analysis of these channels.

Deep-learning models for lipid nanoparticle-based drug delivery.

Background: Early prediction of time-lapse microscopy experiments enables intelligent data management and decision-making. Aim: Using time-lapse data of HepG2 cells exposed to lipid nanoparticles loaded with mRNA for expression of GFP, the authors hypothesized that it is possible to predict in advance whether a cell will express GFP. Methods: The first modeling approach used a convolutional neural network extracting per-cell features at early time points. These features were then combined and explored using either a long short-term memory network (approach 2) or time series feature extraction and gradient boosting machines (approach 3). Results: Accounting for the temporal dynamics significantly improved performance. Conclusion: The results highlight the benefit of accounting for temporal dynamics when studying drug delivery using high-content imaging.

Rapid development of cloud-native intelligent data pipelines for scientific data streams using the HASTE Toolkit.

BACKGROUND: Large streamed datasets, characteristic of life science applications, are often resource-intensive to process, transport and store. We propose a pipeline model, a design pattern for scientific pipelines, where an incoming stream of scientific data is organized into a tiered or ordered "data hierarchy". We introduce the HASTE Toolkit, a proof-of-concept cloud-native software toolkit based on this pipeline model, to partition and prioritize data streams to optimize use of limited computing resources. FINDINGS: In our pipeline model, an "interestingness function" assigns an interestingness score to data objects in the stream, inducing a data hierarchy. From this score, a "policy" guides decisions on how to prioritize computational resource use for a given object. The HASTE Toolkit is a collection of tools to adopt this approach. We evaluate with 2 microscopy imaging case studies. The first is a high content screening experiment, where images are analyzed in an on-premise container cloud to prioritize storage and subsequent computation. The second considers edge processing of images for upload into the public cloud for real-time control of a transmission electron microscope. CONCLUSIONS: Through our evaluation, we created smart data pipelines capable of effective use of storage, compute, and network resources, enabling more efficient data-intensive experiments. We note a beneficial separation between scientific concerns of data priority, and the implementation of this behaviour for different resources in different deployment contexts. The toolkit allows intelligent prioritization to be `bolted on' to new and existing systems - and is intended for use with a range of technologies in different deployment scenarios.

Diagnosis and management of benign biliary strictures post liver transplantation in adults.

Benign biliary strictures after liver transplantation are common and can lead to graft dysfunction and decreased patient survival. Post-transplant strictures are classified as anastomotic or non-anastomotic which differ in response to therapy. Risk factors for biliary strictures following transplantation include impaired blood supply, surgical factors, and biliary anomalies. Patients can present with biliary obstruction but most will be asymptomatic, with only abnormal graft function. MRCP is the most sensitive noninvasive tool for diagnosing biliary complications. In most centres worldwide endoscopy is used first-line in the management of anastomotic strictures, although there is significant variation in endoscopic technique employed; including dilation, placing a single or multiple plastic stents, a fully covered metal stent and most recently using intra-ductal fully covered metal stents. With the introduction of fully covered metal stents the number of interventions patients require has reduced and overall the clinical success of the endoscopic approach has steadily improved. Percutaneous and surgical treatments are now reserved for patients in whom endoscopic management fails or who have had Roux-en-Y anastomoses. However even in these cases, combined procedures with interventional radiology, or implementation of enteroscopy and EUS-guided approaches now means very few patients ultimately require surgical revision.

Deep Learning With Conformal Prediction for Hierarchical Analysis of Large-Scale Whole-Slide Tissue Images.

With the increasing amount of image data collected from biomedical experiments there is an urgent need for smarter and more effective analysis methods. Many scientific questions require analysis of image sub-regions related to some specific biology. Finding such regions of interest (ROIs) at low resolution and limiting the data subjected to final quantification at full resolution can reduce computational requirements and save time. In this paper we propose a three-step pipeline: First, bounding boxes for ROIs are located at low resolution. Next, ROIs are subjected to semantic segmentation into sub-regions at mid-resolution. We also estimate the confidence of the segmented sub-regions. Finally, quantitative measurements are extracted at full resolution. We use deep learning for the first two steps in the pipeline and conformal prediction for confidence assessment. We show that limiting final quantitative analysis to sub-regions with full confidence reduces noise and increases separability of observed biological effects.

Financing urban development, three business models: Johannesburg, Shanghai and London.

There has been growing interest in the expansion of global investment in urban areas, and the financialisation of urban development, both of which bring new business logics into the production of the built environment and shape urban outcomes. At the same time, mega urban projects have continued and spread as a significant format of urban expansion and renewal, often strongly linked to transnational investors and developers. Nonetheless, the distinctive regulatory and political contexts within which transnational actors must bring such projects to fruition matter greatly to outcomes, with territorialised governance arrangements both shaping and being shaped by transnational dynamics. However, there has been little systematic comparative consideration of these diverse regulatory contexts in their own right, rather than as contributors to wider circulating processes such as neoliberalisation. As a result, the implications of different regulatory regimes for urban outcomes have not been effectively assessed. In this paper we therefore broaden the discussion from globalised processes of "financialisation" to consider three large-scale urban development projects from the perspective of their distinctive "business models", including their place in achieving wider strategic objectives at national and metropolitan scales, their agile and often bespoke institutional configurations, and their different forms of financing, taxation and land value capture. Our cases are Lingang, Shanghai (one of nine planned satellite cities), the Corridors of Freedom project in Johannesburg (a linear transport oriented development seeking to integrate the racially divided city), and Old Oak and Park Royal in north-west London (under a mayoral development corporation, associated with significant new metropolitan and national transport investments). We observe that the business models adopted, notably in relation to financial calculations and income streams associated with the developments, are a result of strongly path dependent formats of governance and income generation in each case. However we want to move beyond seeing these as residual, as contingent and contextual to wider accounts of urban development focussed on globalised financial flows and calculations. Using a comparative approach we initiate a systematic analytical conversation about the implications of different business models for the form and socio-economic potential of mega-urban development projects.

Occupancy versus colonization-extinction models for projecting population trends at different spatial scales.

Understanding spatiotemporal population trends and their drivers is a key aim in population ecology. We further need to be able to predict how the dynamics and sizes of populations are affected in the long term by changing landscapes and climate. However, predictions of future population trends are sensitive to a range of modeling assumptions. Deadwood-dependent fungi are an excellent system for testing the performance of different predictive models of sessile species as these species have different rarity and spatial population dynamics, the populations are structured at different spatial scales, and they utilize distinct substrates. We tested how the projected large-scale occupancies of species with differing landscape-scale occupancies are affected over the coming century by different modeling assumptions. We compared projections based on occupancy models against colonization-extinction models, conducting the modeling at alternative spatial scales and using fine- or coarse-resolution deadwood data. We also tested effects of key explanatory variables on species occurrence and colonization-extinction dynamics. The hierarchical Bayesian models applied were fitted to an extensive repeated survey of deadwood and fungi at 174 patches. We projected higher occurrence probabilities and more positive trends using the occupancy models compared to the colonization-extinction models, with greater difference for the species with lower occupancy, colonization rate, and colonization:extinction ratio than for the species with higher estimates of these statistics. The magnitude of future increase in occupancy depended strongly on the spatial modeling scale and resource resolution. We encourage using colonization-extinction models over occupancy models, modeling the process at the finest resource-unit resolution that is utilizable by the species, and conducting projections for the same spatial scale and resource resolution at which the model fitting is conducted. Further, the models applied should include key variables driving the metapopulation dynamics, such as the availability of suitable resource units, habitat quality, and spatial connectivity.

Management of portal cavernoma-associated cholangiopathy: Single-centre experience.

BACKGROUND AND AIMS: Portal cavernoma associated cholangiopathy (PCC) is an uncommon disease in western countries. We describe our experience in seven patients with PCC, in particular the endoscopic management. We describe the mode of presentation, frequent symptoms and the outcome of different treatment modalities of patients with symptomatic PCC. METHODS: Prospectively maintained database was reviewed at a large tertiary referral unit in London, UK. Data included therapeutic interventions, outcomes and complications. RESULTS: Seven patients with PCC were followed for a median of 87 months [interquartile range (IQR), 62-107.5]. Causes of EHPVO included (hypercoagulable status, n=2, peritoneal tuberculosis n=1, neonatal sepsis, n=1, idiopathic, n=3). Acute cholangitis constituted the most recurring complications in all patients during the disease course. Endoscopic intervention was deemed required in all patients for biliary decompression, with 5 out 7 patients managed with repeat endoscopic sessions, (total=23 ERCPs). Surgical portal decompression (meso-caval shunt) was successfully performed in one patient and another patient underwent liver transplantation for decompensated liver cirrhosis. When endoscopic intervention was indicated, a fully covered self expanding metal stent (FcSEMS) provided a longer "symptoms free" period when compared to plastic stent, 7.5 (IQR, 4.75-18.25) and 4 (IQR, 3.5-7) months respectively, P=0.03. Bile duct bleeding occurred in two patients during ERCP procedure, however none of the patients had spontaneous haemobilia. Both patients were successfully treated by FcSEMS. CONCLUSION: Acute cholangitis is a common presentation and recurrent complication during the disease course. Spontaneous haemobilia seems to be uncommon, however it is a significant potential hazard during endoscopic intervention. Insertion of FcSEMS may remodel choledochal varices and provide a longer "symptoms free" period compared to plastic stents.

Transfer Learning with Deep Convolutional Neural Networks for Classifying Cellular Morphological Changes.

The quantification and identification of cellular phenotypes from high-content microscopy images has proven to be very useful for understanding biological activity in response to different drug treatments. The traditional approach has been to use classical image analysis to quantify changes in cell morphology, which requires several nontrivial and independent analysis steps. Recently, convolutional neural networks have emerged as a compelling alternative, offering good predictive performance and the possibility to replace traditional workflows with a single network architecture. In this study, we applied the pretrained deep convolutional neural networks ResNet50, InceptionV3, and InceptionResnetV2 to predict cell mechanisms of action in response to chemical perturbations for two cell profiling datasets from the Broad Bioimage Benchmark Collection. These networks were pretrained on ImageNet, enabling much quicker model training. We obtain higher predictive accuracy than previously reported, between 95% and 97%. The ability to quickly and accurately distinguish between different cell morphologies from a scarce amount of labeled data illustrates the combined benefit of transfer learning and deep convolutional neural networks for interrogating cell-based images.

Deep Learning in Image Cytometry: A Review.

Artificial intelligence, deep convolutional neural networks, and deep learning are all niche terms that are increasingly appearing in scientific presentations as well as in the general media. In this review, we focus on deep learning and how it is applied to microscopy image data of cells and tissue samples. Starting with an analogy to neuroscience, we aim to give the reader an overview of the key concepts of neural networks, and an understanding of how deep learning differs from more classical approaches for extracting information from image data. We aim to increase the understanding of these methods, while highlighting considerations regarding input data requirements, computational resources, challenges, and limitations. We do not provide a full manual for applying these methods to your own data, but rather review previously published articles on deep learning in image cytometry, and guide the readers toward further reading on specific networks and methods, including new methods not yet applied to cytometry data. © 2018 The Authors. Cytometry Part A published by Wiley Periodicals, Inc. on behalf of International Society for Advancement of Cytometry.

Partial diel migration: A facultative migration underpinned by long-term inter-individual variation.

The variations in migration that comprise partial diel migrations, putatively occur entirely as a consequence of behavioural flexibility. However, seasonal partial migrations are increasingly recognised to be mediated by a combination of reversible plasticity in response to environmental variation and individual variation due to genetic and environmental effects. Here, we test the hypothesis that while partial diel migration heterogeneity occurs primarily due to short-term within-individual flexibility in behaviour, long-term individual differences in migratory behaviour also underpin this migration variation. Specifically, we use a hierarchical behavioural reaction norm approach to partition within- and among-individual variation in depth use and diel plasticity in depth use, across short- and long-term time-scales, in a group of 47 burbot (Lota lota) tagged with depth-sensing acoustic telemetry transmitters. We found that within-individual variation at the among-dates-within-seasons and among-seasons scale, explained the dominant proportion of phenotypic variation. However, individuals also repeatedly differed in their expression of migration behaviour over the 2 year study duration. These results reveal that diel migration variation occurs primarily due to short-term within-individual flexibility in depth use and diel migration behaviour. However, repeatable individual differences also played a key role in mediating partial diel migration. These findings represent a significant advancement of our understanding of the mechanisms generating the important, yet poorly understood phenomena of partial diel migration. Moreover, given the pervasive occurrence of diel migrations across aquatic taxa, these findings indicate that individual differences have an important, yet previously unacknowledged role in structuring the temporal and vertical dynamics of aquatic ecosystems.

Forest management could counteract distribution retractions forced by climate change.

Climate change is expected to drive the distribution retraction of northern species. However, particularly in regions with a history of intensive exploitation, changes in habitat management could facilitate distribution expansions counter to expectations under climate change. Here, we test the potential for future forest management to facilitate the southward expansion of an old-forest species from the boreal region into the boreo-nemoral region, contrary to expectations under climate change. We used an ensemble of species distribution models based on citizen science data to project the response of Phellinus ferrugineofuscus, a red-listed old-growth indicator, wood-decaying fungus, to six forest management and climate change scenarios. We projected change in habitat suitability across the boreal and boreo-nemoral regions of Sweden for the period 2020-2100. Scenarios varied in the proportion of forest set aside from production, the level of timber extraction, and the magnitude of climate change. Habitat suitabilities for the study species were projected to show larger relative increases over time in the boreo-nemoral region compared to the boreal region, under all scenarios. By 2100, mean suitabilities in set-aside forest in the boreo-nemoral region were similar to the suitabilities projected for set-aside forest in the boreal region in 2020, suggesting that occurrence in the boreo-nemoral region could be increased. However, across all scenarios, consistently higher projected suitabilities in set-aside forest in the boreal region indicated that the boreal region remained the species stronghold. Furthermore, negative effects of climate change were evident in the boreal region, and projections suggested that climatic changes may eventually counteract the positive effects of forest management in the boreo-nemoral region. Our results suggest that the current rarity of this old-growth indicator species in the boreo-nemoral region may be due to the history of intensive forestry. Forest management therefore has the potential to compensate for the negative effects of climate change. However, increased occurrence at the southern range edge would depend on the dispersal and colonization ability of the species. An increase in the amount of set-aside forest across both the boreal and boreo-nemoral regions is therefore likely to be required to prevent the decline of old-forest species under climate change.

Evaluating citizen science data for forecasting species responses to national forest management.

The extensive spatial and temporal coverage of many citizen science datasets (CSD) makes them appealing for use in species distribution modeling and forecasting. However, a frequent limitation is the inability to validate results. Here, we aim to assess the reliability of CSD for forecasting species occurrence in response to national forest management projections (representing 160,366 km2) by comparison against forecasts from a model based on systematically collected colonization-extinction data. We fitted species distribution models using citizen science observations of an old-forest indicator fungus Phellinus ferrugineofuscus. We applied five modeling approaches (generalized linear model, Poisson process model, Bayesian occupancy model, and two MaxEnt models). Models were used to forecast changes in occurrence in response to national forest management for 2020-2110. Forecasts of species occurrence from models based on CSD were congruent with forecasts made using the colonization-extinction model based on systematically collected data, although different modeling methods indicated different levels of change. All models projected increased occurrence in set-aside forest from 2020 to 2110: the projected increase varied between 125% and 195% among models based on CSD, in comparison with an increase of 129% according to the colonization-extinction model. All but one model based on CSD projected a decline in production forest, which varied between 11% and 49%, compared to a decline of 41% using the colonization-extinction model. All models thus highlighted the importance of protected old forest for P. ferrugineofuscus persistence. We conclude that models based on CSD can reproduce forecasts from models based on systematically collected colonization-extinction data and so lead to the same forest management conclusions. Our results show that the use of a suite of models allows CSD to be reliably applied to land management and conservation decision making, demonstrating that widely available CSD can be a valuable forecasting resource.

Local epiphyte establishment and future metapopulation dynamics in landscapes with different spatiotemporal properties.

Understanding the relative importance of different ecological processes on the metapopulation dynamics of species is the basis for accurately forecasting metapopulation size in fragmented landscapes. Successful local colonization depends on both species dispersal range and how local habitat conditions affect establishment success. Moreover, there is limited understanding of the effects of different spatiotemporal landscape properties on future metapopulation size. We investigate which factors drive the future metapopulation size of the epiphytic model lichen species Lobaria pulmonaria in a managed forest landscape. First, we test the importance of dispersal and local conditions on the colonization-extinction dynamics of the species using Bayesian state-space modelling of a large-scale data set collected over a 10-yr period. Second, we test the importance of dispersal and establishment limitation in explaining establishment probability and subsequent local population growth, based on a 10-yr propagule sowing experiment. Third, we test how future metapopulation size is affected by different metapopulation and spatiotemporal landscape dynamics, using simulations with the metapopulation models fitted to the empirical data. The colonization probability increased with tree inclination and connectivity, with a mean dispersal distance of 97 m (95% credible intervals, 5-530 m). Local extinctions were mainly deterministic set by tree mortality, but also by tree cutting by forestry. No experimental establishments took place on clearcuts, and in closed forest the establishment probability was higher on trees growing on moist than on dry-mesic soils. The subsequent local population growth rate increased with increasing bark roughness. The simulations showed that the restricted dispersal range estimated (compared to non-restricted dispersal range), and short tree rotation length (65 yr instead of 120) had approximately the same negative effects on future metapopulation size, while regeneration of trees creating a random tree pattern instead of an aggregated one had only some negative effect. However, using the colonization rate obtained with the experimentally added diaspores led to a considerable increase in metapopulation size, making the dispersal limitation of the species clear. The future metapopulation size is thus set by the number of host trees located in shady conditions, not isolated from occupied trees, and by the rotation length of these host trees.