Showing papers by "Technical University of Berlin published in 2017"

Quantum-chemical insights from deep tensor neural networks.

Abstract: Learning from data has led to paradigm shifts in a multitude of disciplines, including web, text and image search, speech recognition, as well as bioinformatics. Can machine learning enable similar breakthroughs in understanding quantum many-body systems? Here we develop an efficient deep learning approach that enables spatially and chemically resolved insights into quantum-mechanical observables of molecular systems. We unify concepts from many-body Hamiltonians with purpose-designed deep tensor neural networks, which leads to size-extensive and uniformly accurate (1 kcal mol−1) predictions in compositional and configurational chemical space for molecules of intermediate size. As an example of chemical relevance, the model reveals a classification of aromatic rings with respect to their stability. Further applications of our model for predicting atomic energies and local chemical potentials in molecules, reliable isomer energies, and molecules with peculiar electronic structure demonstrate the potential of machine learning for revealing insights into complex quantum-chemical systems. Machine learning is an increasingly popular approach to analyse data and make predictions. Here the authors develop a ‘deep learning’ framework for quantitative predictions and qualitative understanding of quantum-mechanical observables of chemical systems, beyond properties trivially contained in the training data.

Evaluating the Visualization of What a Deep Neural Network Has Learned

Abstract: Deep neural networks (DNNs) have demonstrated impressive performance in complex machine learning tasks such as image classification or speech recognition. However, due to their multilayer nonlinear structure, they are not transparent, i.e., it is hard to grasp what makes them arrive at a particular classification or recognition decision, given a new unseen data sample. Recently, several approaches have been proposed enabling one to understand and interpret the reasoning embodied in a DNN for a single test image. These methods quantify the “importance” of individual pixels with respect to the classification decision and allow a visualization in terms of a heatmap in pixel/input space. While the usefulness of heatmaps can be judged subjectively by a human, an objective quality measure is missing. In this paper, we present a general methodology based on region perturbation for evaluating ordered collections of pixels such as heatmaps. We compare heatmaps computed by three different methods on the SUN397, ILSVRC2012, and MIT Places data sets. Our main result is that the recently proposed layer-wise relevance propagation algorithm qualitatively and quantitatively provides a better explanation of what made a DNN arrive at a particular classification decision than the sensitivity-based approach or the deconvolution method. We provide theoretical arguments to explain this result and discuss its practical implications. Finally, we investigate the use of heatmaps for unsupervised assessment of the neural network performance.

Understanding activity and selectivity of metal-nitrogen-doped carbon catalysts for electrochemical reduction of CO 2

Abstract: Direct electrochemical reduction of CO2 to fuels and chemicals using renewable electricity has attracted significant attention partly due to the fundamental challenges related to reactivity and selectivity, and partly due to its importance for industrial CO2-consuming gas diffusion cathodes. Here, we present advances in the understanding of trends in the CO2 to CO electrocatalysis of metal- and nitrogen-doped porous carbons containing catalytically active M–N x moieties (M = Mn, Fe, Co, Ni, Cu). We investigate their intrinsic catalytic reactivity, CO turnover frequencies, CO faradaic efficiencies and demonstrate that Fe–N–C and especially Ni–N–C catalysts rival Au- and Ag-based catalysts. We model the catalytically active M–N x moieties using density functional theory and correlate the theoretical binding energies with the experiments to give reactivity-selectivity descriptors. This gives an atomic-scale mechanistic understanding of potential-dependent CO and hydrocarbon selectivity from the M–N x moieties and it provides predictive guidelines for the rational design of selective carbon-based CO2 reduction catalysts. Inexpensive and selective electrocatalysts for CO2 reduction hold promise for sustainable fuel production. Here, the authors report N-coordinated, non-noble metal-doped porous carbons as efficient and selective electrocatalysts for CO2 to CO conversion.

Machine learning of accurate energy-conserving molecular force fields

Abstract: Using conservation of energy-a fundamental property of closed classical and quantum mechanical systems-we develop an efficient gradient-domain machine learning (GDML) approach to construct accurate molecular force fields using a restricted number of samples from ab initio molecular dynamics (AIMD) trajectories. The GDML implementation is able to reproduce global potential energy surfaces of intermediate-sized molecules with an accuracy of 0.3 kcal mol-1 for energies and 1 kcal mol-1 A-1 for atomic forces using only 1000 conformational geometries for training. We demonstrate this accuracy for AIMD trajectories of molecules, including benzene, toluene, naphthalene, ethanol, uracil, and aspirin. The challenge of constructing conservative force fields is accomplished in our work by learning in a Hilbert space of vector-valued functions that obey the law of energy conservation. The GDML approach enables quantitative molecular dynamics simulations for molecules at a fraction of cost of explicit AIMD calculations, thereby allowing the construction of efficient force fields with the accuracy and transferability of high-level ab initio methods.

Electrocatalytic Oxygen Evolution Reaction in Acidic Environments – Reaction Mechanisms and Catalysts

Abstract: The low efficiency of the electrocatalytic oxidation of water to O2 (oxygen evolution reaction-OER) is considered as one of the major roadblocks for the storage of electricity from renewable sources in form of molecular fuels like H2 or hydrocarbons Especially in acidic environments, compatible with the powerful proton exchange membrane (PEM), an earth-abundant OER catalyst that combines high activity and high stability is still unknown Current PEM-compatible OER catalysts still rely mostly on Ir and/or Ru as active components, which are both very scarce elements of the platinum group Hence, the Ir and/or Ru amount in OER catalysts has to be strictly minimized Unfortunately, the OER mechanism, which is the most powerful tool for OER catalyst optimization, still remains unclear In this review, we first summarize the current state of our understanding of the OER mechanism on PEM-compatible heterogeneous electrocatalysts, before we compare and contrast that to the OER mechanism on homogenous catalysts Thereafter, an overview over monometallic OER catalysts is provided to obtain insights into structure-function relations followed by a review of current material optimization concepts and support materials Moreover, missing links required to complete the mechanistic picture as well as the most promising material optimization concepts are pointed out

Future urban land expansion and implications for global croplands.

Abstract: Urban expansion often occurs on croplands. However, there is little scientific understanding of how global patterns of future urban expansion will affect the world's cultivated areas. Here, we combine spatially explicit projections of urban expansion with datasets on global croplands and crop yields. Our results show that urban expansion will result in a 1.8-2.4% loss of global croplands by 2030, with substantial regional disparities. About 80% of global cropland loss from urban expansion will take place in Asia and Africa. In both Asia and Africa, much of the cropland that will be lost is more than twice as productive as national averages. Asia will experience the highest absolute loss in cropland, whereas African countries will experience the highest percentage loss of cropland. Globally, the croplands that are likely to be lost were responsible for 3-4% of worldwide crop production in 2000. Urban expansion is expected to take place on cropland that is 1.77 times more productive than the global average. The loss of cropland is likely to be accompanied by other sustainability risks and threatens livelihoods, with diverging characteristics for different megaurban regions. Governance of urban area expansion thus emerges as a key area for securing livelihoods in the agrarian economies of the Global South.

Mechanism of Lithium Metal Penetration through Inorganic Solid Electrolytes

Abstract: Li deposition is observed and measured on a solid electrolyte in the vicinity of a metallic current collector. Four types of ion-conducting, inorganic solid electrolytes are tested: Amorphous 70/30 mol% Li2S-P2S5, polycrystalline β-Li3PS4, and polycrystalline and single-crystalline Li6La3ZrTaO12 garnet. The nature of lithium plating depends on the proximity of the current collector to defects such as surface cracks and on the current density. Lithium plating penetrates/infiltrates at defects, but only above a critical current density. Eventually, infiltration results in a short circuit between the current collector and the Li-source (anode). These results do not depend on the electrolytes shear modulus and are thus not consistent with the Monroe–Newman model for “dendrites.” The observations suggest that Li-plating in pre-existing flaws produces crack-tip stresses which drive crack propagation, and an electrochemomechanical model of plating-induced Li infiltration is proposed. Lithium short-circuits through solid electrolytes occurs through a fundamentally different process than through liquid electrolytes. The onset of Li infiltration depends on solid-state electrolyte surface morphology, in particular the defect size and density.

Riparian plant litter quality increases with latitude

Abstract: Plant litter represents a major basal resource in streams, where its decomposition is partly regulated by litter traits. Litter-trait variation may determine the latitudinal gradient in decomposition in streams, which is mainly microbial in the tropics and detritivore-mediated at high latitudes. However, this hypothesis remains untested, as we lack information on large-scale trait variation for riparian litter. Variation cannot easily be inferred from existing leaf-trait databases, since nutrient resorption can cause traits of litter and green leaves to diverge. Here we present the first global-scale assessment of riparian litter quality by determining latitudinal variation (spanning 107°) in litter traits (nutrient concentrations; physical and chemical defences) of 151 species from 24 regions and their relationships with environmental factors and phylogeny. We hypothesized that litter quality would increase with latitude (despite variation within regions) and traits would be correlated to produce ‘syndromes’ resulting from phylogeny and environmental variation. We found lower litter quality and higher nitrogen:phosphorus ratios in the tropics. Traits were linked but showed no phylogenetic signal, suggesting that syndromes were environmentally determined. Poorer litter quality and greater phosphorus limitation towards the equator may restrict detritivore-mediated decomposition, contributing to the predominance of microbial decomposers in tropical streams.

SchNet - a deep learning architecture for molecules and materials

Abstract: Deep learning has led to a paradigm shift in artificial intelligence, including web, text and image search, speech recognition, as well as bioinformatics, with growing impact in chemical physics. Machine learning in general and deep learning in particular is ideally suited for representing quantum-mechanical interactions, enabling to model nonlinear potential-energy surfaces or enhancing the exploration of chemical compound space. Here we present the deep learning architecture SchNet that is specifically designed to model atomistic systems by making use of continuous-filter convolutional layers. We demonstrate the capabilities of SchNet by accurately predicting a range of properties across chemical space for \emph{molecules and materials} where our model learns chemically plausible embeddings of atom types across the periodic table. Finally, we employ SchNet to predict potential-energy surfaces and energy-conserving force fields for molecular dynamics simulations of small molecules and perform an exemplary study of the quantum-mechanical properties of C$_{20}$-fullerene that would have been infeasible with regular ab initio molecular dynamics.

The underestimated potential of solar energy to mitigate climate change

Abstract: Despite being currently under-represented in IPCC reports, PV generation represents a growing share of power generation. This Perspective argues that underestimating PV potential led to suboptimal integration measures and that specific deployment strategies for emerging economies should be developed.

Bypassing the Kohn-Sham equations with machine learning

Abstract: Last year, at least 30,000 scientific papers used the Kohn-Sham scheme of density functional theory to solve electronic structure problems in a wide variety of scientific fields. Machine learning holds the promise of learning the energy functional via examples, bypassing the need to solve the Kohn-Sham equations. This should yield substantial savings in computer time, allowing larger systems and/or longer time-scales to be tackled, but attempts to machine-learn this functional have been limited by the need to find its derivative. The present work overcomes this difficulty by directly learning the density-potential and energy-density maps for test systems and various molecules. We perform the first molecular dynamics simulation with a machine-learned density functional on malonaldehyde and are able to capture the intramolecular proton transfer process. Learning density models now allows the construction of accurate density functionals for realistic molecular systems.Machine learning allows electronic structure calculations to access larger system sizes and, in dynamical simulations, longer time scales. Here, the authors perform such a simulation using a machine-learned density functional that avoids direct solution of the Kohn-Sham equations.

Nonribosomal Peptide Synthesis-Principles and Prospects.

Abstract: Nonribosomal peptide synthetases (NRPSs) are large multienzyme machineries that assemble numerous peptides with large structural and functional diversity. These peptides include more than 20 marketed drugs, such as antibacterials (penicillin, vancomycin), antitumor compounds (bleomycin), and immunosuppressants (cyclosporine). Over the past few decades biochemical and structural biology studies have gained mechanistic insights into the highly complex assembly line of nonribosomal peptides. This Review provides state-of-the-art knowledge on the underlying mechanisms of NRPSs and the variety of their products along with detailed analysis of the challenges for future reprogrammed biosynthesis. Such a reprogramming of NRPSs would immediately spur chances to generate analogues of existing drugs or new compound libraries of otherwise nearly inaccessible compound structures.

The Stability Challenges of Oxygen Evolving Catalysts: Towards a Common Fundamental Understanding and Mitigation of Catalyst Degradation.

Abstract: This Review addresses the technical challenges, scientific basis, recent progress, and outlook with respect to the stability and degradation of catalysts for the oxygen evolution reaction (OER) operating at electrolyzer anodes in acidic environments with an emphasis on ion exchange membrane applications. First, the term “catalyst stability” is clarified, as well as current performance targets, major catalyst degradation mechanisms, and their mitigation strategies. Suitable in situ experimental methods are then evaluated to give insight into catalyst degradation and possible pathways to tune OER catalyst stability. Finally, the importance of identifying universal figures of merit for stability is highlighted, leading to a comprehensive accelerated lifetime test that could yield comparable performance data across different laboratories and catalyst types. The aim of this Review is to help disseminate and stress the important relationships between structure, composition, and stability of OER catalysts under different operating conditions.

Connectivity Predicts deep brain stimulation outcome in Parkinson disease.

Abstract: Objective: The benefit of deep brain stimulation (DBS) for Parkinson's disease (PD) may depend on connectivity between the stimulation site and other brain regions, but which regions and whether connectivity can predict outcome in patients remains unknown. Here, we identify the structural and functional connectivity profile of effective DBS to the subthalamic nucleus (STN) and test its ability to predict outcome in an independent cohort. Methods: A training dataset of 51 PD patients with STN DBS was combined with publicly available human connectome data (diffusion tractography and resting state functional connectivity) to identify connections reliably associated with clinical improvement (motor score of Unified Parkinson's Disease Rating Scale). This connectivity profile was then used to predict outcome in an independent cohort of 44 patients from a different center. Results: In the training dataset, connectivity between the DBS electrode and a distributed network of brain regions correlated with clinical response including structural connectivity to supplementary motor area and functional anticorrelation to primary motor cortex (p < 0.001). This same connectivity profile predicted response in an independent patient cohort (p < 0.01). Structural and functional connectivity were independent predictors of clinical improvement (p < 0.001) and estimated response in individual patients with an average error of 15% UPDRS improvement. Results were similar using connectome data from normal subjects or a connectome age, sex, and disease-matched to our DBS patients. Interpretation: Effective STN-DBS for PD is associated with a specific connectivity profile that can predict clinical outcome across independent cohorts. This prediction does not require specialized imaging in PD patients themselves. This article is protected by copyright. All rights reserved.

High-Speed tracking-by-detection without using image information

Abstract: Tracking-by-detection is a common approach to multi-object tracking. With ever increasing performances of object detectors, the basis for a tracker becomes much more reliable. In combination with commonly higher frame rates, this poses a shift in the challenges for a successful tracker. That shift enables the deployment of much simpler tracking algorithms which can compete with more sophisticated approaches at a fraction of the computational cost. We present such an algorithm and show with thorough experiments its potential using a wide range of object detectors. The proposed method can easily run at 100K fps while outperforming the state-of-the-art on the DETRAC vehicle tracking dataset.

The (Un)reliability of saliency methods

Abstract: Saliency methods aim to explain the predictions of deep neural networks. These methods lack reliability when the explanation is sensitive to factors that do not contribute to the model prediction. We use a simple and common pre-processing step which can be compensated for easily—adding a constant shift to the input data—to show that a transformation with no effect on how the model makes the decision can cause numerous methods to attribute incorrectly. In order to guarantee reliability, we believe that the explanation should not change when we can guarantee that two networks process the images in identical manners. We show, through several examples, that saliency methods that do not satisfy this requirement result in misleading attribution. The approach can be seen as a type of unit test; we construct a narrow ground truth to measure one stated desirable property. As such, we hope the community will embrace the development of additional tests.

Nursing skill mix in European hospitals: cross-sectional study of the association with mortality, patient ratings, and quality of care

Abstract: Objectives To determine the association of hospital nursing skill mix with patient mortality, patient ratings of their care and indicators of quality of care. Design Cross-sectional patient discharge data, hospital characteristics and nurse and patient survey data were merged and analysed using generalised estimating equations (GEE) and logistic regression models. Setting Adult acute care hospitals in Belgium, England, Finland, Ireland, Spain and Switzerland. Participants Survey data were collected from 13 077 nurses in 243 hospitals, and 18 828 patients in 182 of the same hospitals in the six countries. Discharge data were obtained for 275 519 surgical patients in 188 of these hospitals. Main outcome measures Patient mortality, patient ratings of care, care quality, patient safety, adverse events and nurse burnout and job dissatisfaction. Results Richer nurse skill mix (eg, every 10-point increase in the percentage of professional nurses among all nursing personnel) was associated with lower odds of mortality (OR=0.89), lower odds of low hospital ratings from patients (OR=0.90) and lower odds of reports of poor quality (OR=0.89), poor safety grades (OR=0.85) and other poor outcomes (0.80

Tracking Catalyst Redox States and Reaction Dynamics in Ni-Fe Oxyhydroxide Oxygen Evolution Reaction Electrocatalysts: The Role of Catalyst Support and Electrolyte pH.

Abstract: Ni–Fe oxyhydroxides are the most active known electrocatalysts for the oxygen evolution reaction (OER) in alkaline electrolytes and are therefore of great scientific and technological importance in the context of electrochemical energy conversion. Here we uncover, investigate, and discuss previously unaddressed effects of conductive supports and the electrolyte pH on the Ni–Fe(OOH) catalyst redox behavior and catalytic OER activity, combining in situ UV–vis spectro-electrochemistry, operando electrochemical mass spectrometry (DEMS), and in situ cryo X-ray absorption spectroscopy (XAS). Supports and pH > 13 strongly enhanced the precatalytic voltammetric charge of the Ni–Fe oxyhydroxide redox peak couple, shifted them more cathodically, and caused a 2–3-fold increase in the catalytic OER activity. Analysis of DEMS-based faradaic oxygen efficiency and electrochemical UV–vis traces consistently confirmed our voltammetric observations, evidencing both a more cathodic O2 release and a more cathodic onset of Ni...

Electrochemical CO2 Reduction: A Classification Problem.

Abstract: In this work we propose four non-coupled binding energies of intermediates as descriptors, or 'genes', for predicting the product distribution in CO2 electroreduction. Simple reactions can be understood by the Sabatier principle (catalytic activity vs. one descriptor), while more complex reactions tend to give multiple very different products and consequently the product selectivity is a more complex property to understand. We approach this, as a logistical classification problem, by grouping metals according to their major experimental product from CO2 electroreduction: H2, CO, formic acid and beyond CO* (hydrocarbons or alcohols). We compare the groups in terms of multiple binding energies of intermediates calculated by density functional theory. Here we find three descriptors to explain the grouping: the adsorption energies of H*, COOH* and CO*. To further classify products beyond CO*, we carry out formaldehyde experiments on Cu, Ag and Au and combine these results with the literature to group and differentiate alcohol or hydrocarbon products. We find that the oxygen binding (adsorption energy of CH3O*) is an additional descriptor to explain the alcohol formation in reduction processes. Finally, the adsorption energy of the four intermediates, H*, COOH*, CO* and CH3O*, can be used to differentiate, group and explain products in electrochemical reduction processes involving CO2, CO and carbon-oxygen compounds.

A New Paradigm for Systematic Literature Reviews in Supply Chain Management

Abstract: While systematic literature reviews (SLRs) have contributed substantially to developing knowledge in fields such as medicine, they have made limited contributions to developing knowledge in the supply chain management domain. This is due to the ontological and epistemological idiosyncrasies of research in supply chain management, which need to be accounted for when retrieving, selecting, and synthesizing studies in an SLR. Therefore, we propose a new paradigm for SLRs in the supply chain domain that is based on both best practice and the unique attributes of doing supply chain management research. This approach involves exploring existing studies with attention to theoretical boundaries, units of analysis, sources of data, study contexts, and definitions and the operationalization of constructs, as well as research methods, with the goal of refining or revising existing theory. This new paradigm will push supply chain management research to the frontier of current methodological standards and build a foundation for improving the contribution of future SLRs in the supply chain and adjacent management disciplines.

The Slashdot Zoo: Mining a Social Network with Negative Edges

Abstract: We analyse the corpus of user relationships of the Slashdot technology news site. The data was collected from the Slashdot Zoo feature where users of the website can tag other users as friends and foes, providing positive and negative endorsements. We adapt social network analysis techniques to the problem of negative edge weights. In particular, we consider signed variants of global network characteristics such as the clustering coefficient, node-level characteristics such as centrality and popularity measures, and link-level characteristics such as distances and similarity measures. We evaluate these measures on the task of identifying unpopular users, as well as on the task of predicting the sign of links and show that the network exhibits multiplicative transitivity which allows algebraic methods based on matrix multiplication to be used. We compare our methods to traditional methods which are only suitable for positively weighted edges.

Functional Graphene Nanomaterials Based Architectures: Biointeractions, Fabrications, and Emerging Biological Applications

Abstract: Functional graphene nanomaterials (FGNs) are fast emerging materials with extremely unique physical and chemical properties and physiological ability to interfere and/or interact with bioorganisms; as a result, FGNs present manifold possibilities for diverse biological applications. Beyond their use in drug/gene delivery, phototherapy, and bioimaging, recent studies have revealed that FGNs can significantly promote interfacial biointeractions, in particular, with proteins, mammalian cells/stem cells, and microbials. FGNs can adsorb and concentrate nutrition factors including proteins from physiological media. This accelerates the formation of extracellular matrix, which eventually promotes cell colonization by providing a more beneficial microenvironment for cell adhesion and growth. Furthermore, FGNs can also interact with cocultured cells by physical or chemical stimulation, which significantly mediate their cellular signaling and biological performance. In this review, we elucidate FGNs–bioorganism int...

Fossil-fueled development (SSP5): An energy and resource intensive scenario for the 21st century

Abstract: This paper presents a set of energy and resource intensive scenarios based on the concept of Shared Socio-Economic Pathways (SSPs). The scenario family is characterized by rapid and fossil-fueled development with high socio-economic challenges to mitigation and low socio-economic challenges to adaptation (SSP5). A special focus is placed on the SSP5 marker scenario developed by the REMIND-MAgPIE integrated assessment modeling framework. The SSP5 baseline scenarios exhibit very high levels of fossil fuel use, up to a doubling of global food demand, and up to a tripling of energy demand and greenhouse gas emissions over the course of the century, marking the upper end of the scenario literature in several dimensions. These scenarios are currently the only SSP scenarios that result in a radiative forcing pathway as high as the highest Representative Concentration Pathway (RCP8.5). This paper further investigates the direct impact of mitigation policies on the SSP5 energy, land and emissions dynamics confirming high socio-economic challenges to mitigation in SSP5. Nonetheless, mitigation policies reaching climate forcing levels as low as in the lowest Representative Concentration Pathway (RCP2.6) are accessible in SSP5. The SSP5 scenarios presented in this paper aim to provide useful reference points for future climate change, climate impact, adaption and mitigation analysis, and broader questions of sustainable development.

Synthetic chemicals as agents of global change

Abstract: Though concerns about the proliferation of synthetic chemicals – including pesticides – gave rise to the modern environmental movement in the early 1960s, synthetic chemical pollution has not been included in most analyses of global change. We examined the rate of change in the production and variety of pesticides, pharmaceuticals, and other synthetic chemicals over the past four decades. We compared these rates to those for well-recognized drivers of global change such as rising atmospheric CO2 concentrations, nutrient pollution, habitat destruction, and biodiversity loss. Our analysis showed that increases in synthetic chemical production and diversification, particularly within the developing world, outpaced these other agents of global change. Despite these trends, mainstream ecological journals, ecological meetings, and ecological funding through the US National Science Foundation devote less than 2% of their journal pages, meeting talks, and science funding, respectively, to the study of synthetic chemicals.

A synthesis of radial growth patterns preceding tree mortality

Abstract: Tree mortality is a key factor influencing forest functions and dynamics, but our understanding of the mechanisms leading to mortality and the associated changes in tree growth rates are still limited. We compiled a new pan-conti- nental tree-ring width database from sites where both dead and living trees were sampled (2970 dead and 4224 living trees from 190 sites, including 36 species), and compared early and recent growth rates between trees that died and those that survived a given mortality event. We observed a decrease in radial growth before death in ca. 84% of the mortality events. The extent and duration of these reductions were highly variable (1–100 years in 96% of events) due to the complex interactions among study species and the source(s) of mortality. Strong and long-lasting declines were found for gymnosperms, shade- and drought-tolerant species, and trees that died from competition. Angiosperms and trees that died due to biotic attacks (especially bark-beetles) typically showed relatively small and short-term growth reductions. Our analysis did not highlight any universal trade-off between early growth and tree longevity within a species, although this result may also reflect high variability in sampling design among sites. The intersite and interspecific variability in growth patterns before mortality provides valuable information on the nature of the mortality process, which is consistent with our understanding of the physiological mechanisms leading to mortality. Abrupt changes in growth immediately before death can be associated with generalized hydraulic failure and/or bark-beetle attack, while long-term decrease in growth may be associated with a gradual decline in hydraulic performance coupled with depletion in carbon reserves. Our results imply that growth-based mortality algorithms may be a powerful tool for predicting gymnosperm mortality induced by chronic stress, but not necessarily so for angiosperms and in case of intense drought or bark-beetle outbreaks.

Comparative genomics reveals high biological diversity and specific adaptations in the industrially and medically important fungal genus Aspergillus

Abstract: Background: The fungal genus Aspergillus is of critical importance to humankind. Species include those with industrial applications, important pathogens of humans, animals and crops, a source of potent carcinogenic contaminants of food, and an important genetic model. The genome sequences of eight aspergilli have already been explored to investigate aspects of fungal biology, raising questions about evolution and specialization within this genus. Results: We have generated genome sequences for ten novel, highly diverse Aspergillus species and compared these in detail to sister and more distant genera. Comparative studies of key aspects of fungal biology, including primary and secondary metabolism, stress response, biomass degradation, and signal transduction, revealed both conservation and diversity among the species. Observed genomic differences were validated with experimental studies. This revealed several highlights, such as the potential for sex in asexual species, organic acid production genes being a key feature of black aspergilli, alternative approaches for degrading plant biomass, and indications for the genetic basis of stress response. A genome-wide phylogenetic analysis demonstrated in detail the relationship of the newly genome sequenced species with other aspergilli. Conclusions: Many aspects of biological differences between fungal species cannot be explained by current knowledge obtained from genome sequences. The comparative genomics and experimental study, presented here, allows for the first time a genus-wide view of the biological diversity of the aspergilli and in many, but not all, cases linked genome differences to phenotype. Insights gained could be exploited for biotechnological and medical applications of fungi.