Guideline-based and bioinformatic reassessment of lesion-associated gene and variant pathogenicity in focal human epilepsies

in Epilepsia (2018)

Objective: Increasing availability of surgically resected brain tissue from patients with focal epilepsy and Focal Cortical Dysplasia (FCD) or low-grade glio-neuronal tumors has fostered large-scale ... [more ▼]

Objective: Increasing availability of surgically resected brain tissue from patients with focal epilepsy and Focal Cortical Dysplasia (FCD) or low-grade glio-neuronal tumors has fostered large-scale genetic examination. However, assessment of pathogenicity of germline and somatic variants remains difficult. Here, we present a state of the art evaluation of reported genes and variants associated with epileptic brain lesions. Methods: We critically re-evaluated the pathogenicity for all neuropathology-associated variants reported to date in PubMed and ClinVar databases including 101 neuropathology-associated missense variants encompassing 11 disease-related genes. We assessed gene variant tolerance and classified all identified missense variants according to guidelines from the American College of Medical Genetics and Genomics (ACMG). We further extended the bioinformatic variant prediction by introducing a novel gene-specific deleteriousness ranking for prediction scores. Results: Application of ACMG guidelines and in silico gene variant tolerance analysis classified only seven out of 11 genes to be likely disease-associated according to the reported a disease mechanism, while 61 (60.4%) of 101 variants of those genes were classified as of uncertain significance (VUS), 37 (36.6%) as being likely pathogenic (LP) and 3 (3%) as being pathogenic (P). Significance: We concluded that the majority of neuropathology-associated variants reported to date do not have enough evidence to be classified as pathogenic. Interpretation of lesion-associated variants remains challenging and application of current ACMG guidelines is recommended for interpretation and prediction. [less ▲]

Ergebnisse der Zukunftswerkstatt im Rahmen des Praxis- und Kontakttages 2018

Article for general public (2018)

Im Rahmen des vom PraxisBüro der Universität Luxemburg organisierten Praxis- und Kontakttages 2018, haben ca. 300 Professionelle der Sozialen Arbeit an einer Zukunftswerkstatt mit dem Thema "Was sind die ... [more ▼]

Im Rahmen des vom PraxisBüro der Universität Luxemburg organisierten Praxis- und Kontakttages 2018, haben ca. 300 Professionelle der Sozialen Arbeit an einer Zukunftswerkstatt mit dem Thema "Was sind die Bedürfnisse der Akteure der Sozialen Arbeit?" teilgenommen. Die Ergebnisse wurden gesammelt und von der Autorin analysiert. Der Artikel beschreibt die Ergebnisse der Zukunftswerkstatt sowie ihre Bedeutung für die Soziale Arbeit in Luxemburg. [less ▲]

Privacy-preserving Recommender Systems Facilitated By The Machine Learning Approach

Doctoral thesis (2018)

Recommender systems, which play a critical role in e-business services, are closely linked to our daily life. For example, companies such as Youtube and Amazon are always trying to secure their profit by ... [more ▼]

Recommender systems, which play a critical role in e-business services, are closely linked to our daily life. For example, companies such as Youtube and Amazon are always trying to secure their profit by estimating personalized user preferences and recommending the most relevant items (e.g., products, news, etc.) to each user from a large number of candidates. State-of-the-art recommender systems are often built on top of collaborative filtering techniques, of which the accuracy performance relies on precisely modeling user-item interactions by analyzing massive user historical data, such as browsing history, purchasing records, locations and so on. Generally, more data can lead to more accurate estimations and more commercial strategies, as such, service providers have incentives to collect and use more user data. On the one hand, recommender systems bring more income to service providers and more convenience to users; on the other hand, the user data can be abused, arising immediate privacy risks to the public. Therefore, how to preserve privacy while enjoying recommendation services becomes an increasingly important topic to both the research community and commercial practitioners. The privacy concerns can be disparate when constructing recommender systems or providing recommendation services under different scenarios. One scenario is that, a service provider wishes to protect its data privacy from the inference attack, a technique aims to infer more information (e.g., whether a record is in or not) about a database, by analyzing statistical outputs; the other scenario is that, multiple users agree to jointly perform a recommendation task, but none of them is willing to share their private data with any other users. Security primitives, such as homomorphic encryption, secure multiparty computation, and differential privacy, are immediate candidates to address privacy concerns. A typical approach to build efficient and accurate privacy-preserving solutions is to improve the security primitives, and then apply them to existing recommendation algorithms. However, this approach often yields a solution far from the satisfactory-of-practice, as most users have a low tolerance to the latency-increase or accuracy-drop, regarding recommendation services. The PhD program explores machine learning aided approaches to build efficient privacy-preserving solutions for recommender systems. The results of each proposed solution demonstrate that machine learning can be a strong assistant for privacy-preserving, rather than only a troublemaker. [less ▲]

Mit schimmernder Rüstung und flatterndem Cape. Aventiuren zwischen Mittelalter und Moderne

Speeches/Talks (2018)

The long and winding road to inclusion in Higher Education

Scientific Conference (2018, October 18)

Wine as an Investment

Doctoral thesis (2018)

During the last two decades, the wine auction market experienced non-monotonic dynamics and provided abnormal returns. When fine wine prices started to rise in the mid-1980s, wine investing transcended ... [more ▼]

During the last two decades, the wine auction market experienced non-monotonic dynamics and provided abnormal returns. When fine wine prices started to rise in the mid-1980s, wine investing transcended from a pastime into a serious investment transforming wine into a widespread investment activity. As an alternative asset, fine wine attracts increased attention among individual and institutional investors in recent years, particularly during the financial crisis due to its diversification potential. Economists also began evaluating wine as a new alternative asset class. In this thesis, we review the wine auction markets, study the price dynamics, and answer some fundamental questions. To build solid conclusions, we developed the largest French wine auction database encompassing the last 20 years. We summary our main contributions as follows. First, we review and explain the growth in recent global wine auction markets. Second, we construct price indices with hedonic regression models to observe market movements. Third, we investigate wine investment’s diversification potential. Wine in an optimal portfolio can improve the risk-return characteristics; however, the augmentation depends significantly on the market segmentation to which the investor applies it. Fourth, we analyze the wine price bubbles and collapses detection using a newly developed econometric approach. We elucidate strong evidence of two bubbles in the Bordeaux and Burgundy wine auction markets, whereas Rhône wine price behaved in a similar, but less significant, trend as Bordeaux and Burgundy wine markets. Finally, we examine the price determinants with a particular focus on the expert effect on this experienced good. We illuminate expert influence remains economically and statistically significant throughout the sample period. Using event study methodology, we disentangle the expert effect from other price variations and assert the market significantly reflects expert’s big re-ratings in the short term, yet the effect diminishes over time. However, this re-rating effect holds insignificant for modest changes. [less ▲]

The provision of urban green space and its accessibility: Spatial data effects in Brussels

in PLoS ONE (2018), 13(10), 0204684

Urban green space (UGS) has many environmental and social benefits. UGS provision and access are increasingly considered in urban policies and must rely on data and indicators that can capture variations ... [more ▼]

Urban green space (UGS) has many environmental and social benefits. UGS provision and access are increasingly considered in urban policies and must rely on data and indicators that can capture variations in the distribution of UGS within cities. There is no consensus about how UGS, and their provision and access, must be defined from different land use data types. Here we identify four spatial dimensions of UGS and critically examine how different data sources affect these dimensions and our understanding of their variation within a city region (Brussels). We compare UGS indicators measured from an imagery source (NDVI from Landsat), an official cadastre-based map, and the voluntary geographical information provided by OpenStreetMap (OSM). We compare aggregate values of provision and access to UGS as well as their spatial distribution along a centrality gradient and at neighbourhood scale. We find that there are strong differences in the value of indicators when using the different datasets, especially due to their ability to capture private and public green space. However we find that the interpretation of intra-urban spatial variations is not affected by changes in data source. Centrality in particular is a strong determinant of the relative values of UGS availability, fragmentation and accessibility, irrespective of datasets. [less ▲]

Towards exact molecular dynamics simulations with machine-learned force fields

in Nature Communications (2018), 9

Molecular dynamics (MD) simulations employing classical force fields constitute the cornerstone of contemporary atomistic modeling in chemistry, biology, and materials science. However, the predictive ... [more ▼]

Molecular dynamics (MD) simulations employing classical force fields constitute the cornerstone of contemporary atomistic modeling in chemistry, biology, and materials science. However, the predictive power of these simulations is only as good as the underlying interatomic potential. Classical potentials often fail to faithfully capture key quantum effects in molecules and materials. Here we enable the direct construction of flexible molecular force fields from high-level ab initio calculations by incorporating spatial and temporal physical symmetries into a gradient-domain machine learning (sGDML) model in an automatic data-driven way. The developed sGDML approach faithfully reproduces global force fields at quantum-chemical CCSD(T) level of accuracy and allows converged molecular dynamics simulations with fully quantized electrons and nuclei. We present MD simulations, for flexible molecules with up to a few dozen atoms and provide insights into the dynamical behavior of these molecules. Our approach provides the key missing ingredient for achieving spectroscopic accuracy in molecular simulations. [less ▲]

Shortcuts to adiabaticity in Fermi gases

in New Journal of Physics (2018)

̒'Enforcement/Execution' of ICSID Awards Against Reluctant States

in Journal of World Investment and Trade (2018)