@article{11432, abstract = {This paper proposes a method for simulating liquids in large bodies of water by coupling together a water surface wave simulator with a 3D Navier-Stokes simulator. The surface wave simulation uses the equivalent sources method (ESM) to efficiently animate large bodies of water with precisely controllable wave propagation behavior. The 3D liquid simulator animates complex non-linear fluid behaviors like splashes and breaking waves using off-the-shelf simulators using FLIP or the level set method with semi-Lagrangian advection. We combine the two approaches by using the 3D solver to animate localized non-linear behaviors, and the 2D wave solver to animate larger regions with linear surface physics. We use the surface motion from the 3D solver as boundary conditions for 2D surface wave simulator, and we use the velocity and surface heights from the 2D surface wave simulator as boundary conditions for the 3D fluid simulation. We also introduce a novel technique for removing visual artifacts caused by numerical errors in 3D fluid solvers: we use experimental data to estimate the artificial dispersion caused by the 3D solver and we then carefully tune the wave speeds of the 2D solver to match it, effectively eliminating any differences in wave behavior across the boundary. To the best of our knowledge, this is the first time such a empirically driven error compensation approach has been used to remove coupling errors from a physics simulator. Our coupled simulation approach leverages the strengths of each simulation technique, animating large environments with seamless transitions between 2D and 3D physics.}, author = {Schreck, Camille and Wojtan, Christopher J}, issn = {1467-8659}, journal = {Computer Graphics Forum}, number = {2}, pages = {343--353}, publisher = {Wiley}, title = {{Coupling 3D liquid simulation with 2D wave propagation for large scale water surface animation using the equivalent sources method}}, doi = {10.1111/cgf.14478}, volume = {41}, year = {2022}, } @article{11469, abstract = {Thermalizing and localized many-body quantum systems present two distinct dynamical phases of matter. Recently the fate of a localized system coupled to a thermalizing system viewed as a quantum bath received significant theoretical and experimental attention. In this work, we study a mobile impurity, representing a small quantum bath, that interacts locally with an Anderson insulator with a finite density of localized particles. Using static Hartree approximation to obtain an effective disorder strength, we formulate an analytic criterion for the perturbative stability of the localization. Next, we use an approximate dynamical Hartree method and the quasi-exact time-evolved block decimation (TEBD) algorithm to study the dynamics of the system. We find that the dynamical Hartree approach which completely ignores entanglement between the impurity and localized particles predicts the delocalization of the system. In contrast, the full numerical simulation of the unitary dynamics with TEBD suggests the stability of localization on numerically accessible timescales. Finally, using an extension of the density matrix renormalization group algorithm to excited states (DMRG-X), we approximate the highly excited eigenstates of the system. We find that the impurity remains localized in the eigenstates and entanglement is enhanced in a finite region around the position of the impurity, confirming the dynamical predictions. Dynamics and the DMRG-X results provide compelling evidence for the stability of localization.}, author = {Brighi, Pietro and Michailidis, Alexios and Kirova, Kristina and Abanin, Dmitry A. and Serbyn, Maksym}, issn = {2469-9969}, journal = {Physical Review B}, number = {22}, publisher = {American Physical Society}, title = {{Localization of a mobile impurity interacting with an Anderson insulator}}, doi = {10.1103/physrevb.105.224208}, volume = {105}, year = {2022}, } @article{11470, abstract = {Many-body localization (MBL) is an example of a dynamical phase of matter that avoids thermalization. While the MBL phase is robust to weak local perturbations, the fate of an MBL system coupled to a thermalizing quantum system that represents a “heat bath” is an open question that is actively investigated theoretically and experimentally. In this work, we consider the stability of an Anderson insulator with a finite density of particles interacting with a single mobile impurity—a small quantum bath. We give perturbative arguments that support the stability of localization in the strong interaction regime. Large-scale tensor network simulations of dynamics are employed to corroborate the presence of the localized phase and give quantitative predictions in the thermodynamic limit. We develop a phenomenological description of the dynamics in the strong interaction regime, and we demonstrate that the impurity effectively turns the Anderson insulator into an MBL phase, giving rise to nontrivial entanglement dynamics well captured by our phenomenology.}, author = {Brighi, Pietro and Michailidis, Alexios A. and Abanin, Dmitry A. and Serbyn, Maksym}, issn = {2469-9969}, journal = {Physical Review B}, number = {22}, publisher = {American Physical Society}, title = {{Propagation of many-body localization in an Anderson insulator}}, doi = {10.1103/physrevb.105.l220203}, volume = {105}, year = {2022}, } @article{11736, abstract = {This paper introduces a methodology for inverse-modeling of yarn-level mechanics of cloth, based on the mechanical response of fabrics in the real world. We compiled a database from physical tests of several different knitted fabrics used in the textile industry. These data span different types of complex knit patterns, yarn compositions, and fabric finishes, and the results demonstrate diverse physical properties like stiffness, nonlinearity, and anisotropy. We then develop a system for approximating these mechanical responses with yarn-level cloth simulation. To do so, we introduce an efficient pipeline for converting between fabric-level data and yarn-level simulation, including a novel swatch-level approximation for speeding up computation, and some small-but-necessary extensions to yarn-level models used in computer graphics. The dataset used for this paper can be found at http://mslab.es/projects/YarnLevelFabrics.}, author = {Sperl, Georg and Sánchez-Banderas, Rosa M. and Li, Manwen and Wojtan, Christopher J and Otaduy, Miguel A.}, issn = {1557-7368}, journal = {ACM Transactions on Graphics}, number = {4}, publisher = {Association for Computing Machinery}, title = {{Estimation of yarn-level simulation models for production fabrics}}, doi = {10.1145/3528223.3530167}, volume = {41}, year = {2022}, } @article{12109, abstract = {Kelvin probe force microscopy (KPFM) is a powerful tool for studying contact electrification (CE) at the nanoscale, but converting KPFM voltage maps to charge density maps is nontrivial due to long-range forces and complex system geometry. Here we present a strategy using finite-element method (FEM) simulations to determine the Green's function of the KPFM probe/insulator/ground system, which allows us to quantitatively extract surface charge. Testing our approach with synthetic data, we find that accounting for the atomic force microscope (AFM) tip, cone, and cantilever is necessary to recover a known input and that existing methods lead to gross miscalculation or even the incorrect sign of the underlying charge. Applying it to experimental data, we demonstrate its capacity to extract realistic surface charge densities and fine details from contact-charged surfaces. Our method gives a straightforward recipe to convert qualitative KPFM voltage data into quantitative charge data over a range of experimental conditions, enabling quantitative CE at the nanoscale.}, author = {Pertl, Felix and Sobarzo Ponce, Juan Carlos A and Shafeek, Lubuna B and Cramer, Tobias and Waitukaitis, Scott R}, issn = {2475-9953}, journal = {Physical Review Materials}, number = {12}, publisher = {American Physical Society}, title = {{Quantifying nanoscale charge density features of contact-charged surfaces with an FEM/KPFM-hybrid approach}}, doi = {10.1103/PhysRevMaterials.6.125605}, volume = {6}, year = {2022}, } @article{12142, abstract = {Theory for liability-scale models of the underlying genetic basis of complex disease provides an important way to interpret, compare, and understand results generated from biological studies. In particular, through estimation of the liability-scale heritability (LSH), liability models facilitate an understanding and comparison of the relative importance of genetic and environmental risk factors that shape different clinically important disease outcomes. Increasingly, large-scale biobank studies that link genetic information to electronic health records, containing hundreds of disease diagnosis indicators that mostly occur infrequently within the sample, are becoming available. Here, we propose an extension of the existing liability-scale model theory suitable for estimating LSH in biobank studies of low-prevalence disease. In a simulation study, we find that our derived expression yields lower mean square error (MSE) and is less sensitive to prevalence misspecification as compared to previous transformations for diseases with =< 2% population prevalence and LSH of =< 0.45, especially if the biobank sample prevalence is less than that of the wider population. Applying our expression to 13 diagnostic outcomes of =< 3% prevalence in the UK Biobank study revealed important differences in LSH obtained from the different theoretical expressions that impact the conclusions made when comparing LSH across disease outcomes. This demonstrates the importance of careful consideration for estimation and prediction of low-prevalence disease outcomes and facilitates improved inference of the underlying genetic basis of =< 2% population prevalence diseases, especially where biobank sample ascertainment results in a healthier sample population.}, author = {Ojavee, Sven E. and Kutalik, Zoltan and Robinson, Matthew Richard}, issn = {0002-9297}, journal = {The American Journal of Human Genetics}, keywords = {Genetics (clinical), Genetics}, number = {11}, pages = {2009--2017}, publisher = {Elsevier}, title = {{Liability-scale heritability estimation for biobank studies of low-prevalence disease}}, doi = {10.1016/j.ajhg.2022.09.011}, volume = {109}, year = {2022}, } @article{12495, abstract = {Fairness-aware learning aims at constructing classifiers that not only make accurate predictions, but also do not discriminate against specific groups. It is a fast-growing area of machine learning with far-reaching societal impact. However, existing fair learning methods are vulnerable to accidental or malicious artifacts in the training data, which can cause them to unknowingly produce unfair classifiers. In this work we address the problem of fair learning from unreliable training data in the robust multisource setting, where the available training data comes from multiple sources, a fraction of which might not be representative of the true data distribution. We introduce FLEA, a filtering-based algorithm that identifies and suppresses those data sources that would have a negative impact on fairness or accuracy if they were used for training. As such, FLEA is not a replacement of prior fairness-aware learning methods but rather an augmentation that makes any of them robust against unreliable training data. We show the effectiveness of our approach by a diverse range of experiments on multiple datasets. Additionally, we prove formally that –given enough data– FLEA protects the learner against corruptions as long as the fraction of affected data sources is less than half. Our source code and documentation are available at https://github.com/ISTAustria-CVML/FLEA.}, author = {Iofinova, Eugenia B and Konstantinov, Nikola H and Lampert, Christoph}, issn = {2835-8856}, journal = {Transactions on Machine Learning Research}, publisher = {ML Research Press}, title = {{FLEA: Provably robust fair multisource learning from unreliable training data}}, year = {2022}, } @article{10284, abstract = {Infections early in life can have enduring effects on an organism's development and immunity. In this study, we show that this equally applies to developing ‘superorganisms’––incipient social insect colonies. When we exposed newly mated Lasius niger ant queens to a low pathogen dose, their colonies grew more slowly than controls before winter, but reached similar sizes afterwards. Independent of exposure, queen hibernation survival improved when the ratio of pupae to workers was small. Queens that reared fewer pupae before worker emergence exhibited lower pathogen levels, indicating that high brood rearing efforts interfere with the ability of the queen's immune system to suppress pathogen proliferation. Early-life queen pathogen exposure also improved the immunocompetence of her worker offspring, as demonstrated by challenging the workers to the same pathogen a year later. Transgenerational transfer of the queen's pathogen experience to her workforce can hence durably reduce the disease susceptibility of the whole superorganism.}, author = {Casillas Perez, Barbara E and Pull, Christopher and Naiser, Filip and Naderlinger, Elisabeth and Matas, Jiri and Cremer, Sylvia}, issn = {1461-0248}, journal = {Ecology Letters}, number = {1}, pages = {89--100}, publisher = {Wiley}, title = {{Early queen infection shapes developmental dynamics and induces long-term disease protection in incipient ant colonies}}, doi = {10.1111/ele.13907}, volume = {25}, year = {2022}, } @inproceedings{10752, abstract = {The digitalization of almost all aspects of our everyday lives has led to unprecedented amounts of data being freely available on the Internet. In particular social media platforms provide rich sources of user-generated data, though typically in unstructured form, and with high diversity, such as written in many different languages. Automatically identifying meaningful information in such big data resources and extracting it efficiently is one of the ongoing challenges of our time. A common step for this is sentiment analysis, which forms the foundation for tasks such as opinion mining or trend prediction. Unfortunately, publicly available tools for this task are almost exclusively available for English-language texts. Consequently, a large fraction of the Internet users, who do not communicate in English, are ignored in automatized studies, a phenomenon called rare-language discrimination.In this work we propose a technique to overcome this problem by a truly multi-lingual model, which can be trained automatically without linguistic knowledge or even the ability to read the many target languages. The main step is to combine self-annotation, specifically the use of emoticons as a proxy for labels, with multi-lingual sentence representations.To evaluate our method we curated several large datasets from data obtained via the free Twitter streaming API. The results show that our proposed multi-lingual training is able to achieve sentiment predictions at the same quality level for rare languages as for frequent ones, and in particular clearly better than what mono-lingual training achieves on the same data. }, author = {Lampert, Jasmin and Lampert, Christoph}, booktitle = {2021 IEEE International Conference on Big Data}, isbn = {9781665439022}, location = {Orlando, FL, United States}, pages = {5185--5192}, publisher = {IEEE}, title = {{Overcoming rare-language discrimination in multi-lingual sentiment analysis}}, doi = {10.1109/bigdata52589.2021.9672003}, year = {2022}, } @article{12244, abstract = {Environmental cues influence the highly dynamic morphology of microglia. Strategies to characterize these changes usually involve user-selected morphometric features, which preclude the identification of a spectrum of context-dependent morphological phenotypes. Here we develop MorphOMICs, a topological data analysis approach, which enables semiautomatic mapping of microglial morphology into an atlas of cue-dependent phenotypes and overcomes feature-selection biases and biological variability. We extract spatially heterogeneous and sexually dimorphic morphological phenotypes for seven adult mouse brain regions. This sex-specific phenotype declines with maturation but increases over the disease trajectories in two neurodegeneration mouse models, with females showing a faster morphological shift in affected brain regions. Remarkably, microglia morphologies reflect an adaptation upon repeated exposure to ketamine anesthesia and do not recover to control morphologies. Finally, we demonstrate that both long primary processes and short terminal processes provide distinct insights to morphological phenotypes. MorphOMICs opens a new perspective to characterize microglial morphology.}, author = {Colombo, Gloria and Cubero, Ryan J and Kanari, Lida and Venturino, Alessandro and Schulz, Rouven and Scolamiero, Martina and Agerberg, Jens and Mathys, Hansruedi and Tsai, Li-Huei and Chachólski, Wojciech and Hess, Kathryn and Siegert, Sandra}, issn = {1546-1726}, journal = {Nature Neuroscience}, keywords = {General Neuroscience}, number = {10}, pages = {1379--1393}, publisher = {Springer Nature}, title = {{A tool for mapping microglial morphology, morphOMICs, reveals brain-region and sex-dependent phenotypes}}, doi = {10.1038/s41593-022-01167-6}, volume = {25}, year = {2022}, } @phdthesis{12378, abstract = {Environmental cues influence the highly dynamic morphology of microglia. Strategies to characterize these changes usually involve user-selected morphometric features, which preclude the identification of a spectrum of context-dependent morphological phenotypes. Here, we develop MorphOMICs, a topological data analysis approach, which enables semiautomatic mapping of microglial morphology into an atlas of cue-dependent phenotypes, overcomes feature-selection bias and minimizes biological variability. First, with MorphOMICs we derive the morphological spectrum of microglia across seven brain regions during postnatal development and in two distinct Alzheimer’s disease degeneration mouse models. We uncover region-specific and sexually dimorphic morphological trajectories, with females showing an earlier morphological shift than males in the degenerating brain. Overall, we demonstrate that both long primary- and short terminal processes provide distinct insights to morphological phenotypes. Moreover, using machine learning to map novel condition on the spectrum, we observe that microglia morphologies reflect a dose-dependent adaptation upon ketamine anesthesia and do not recover to control morphologies. Next, we took advantage of MorphOMICs to build a high-resolution and layer-specific map of microglial morphological spectrum in the retina, covering postnatal development and rd10 degeneration. Here, following photoreceptor death, microglia assume an early developmentlike morphology. Finally, we map microglial morphology following optic nerve crush on the retinal spectrum and observe a layer- and sex-dependent response. Overall, MorphOMICs opens a new perspective to analyze microglial morphology across multiple conditions, and provides a novel tool to characterize microglial morphology beyond the traditionally dichotomized view of microglia.}, author = {Colombo, Gloria}, issn = {2663-337X}, pages = {142}, publisher = {Institute of Science and Technology Austria}, title = {{MorphOMICs, a tool for mapping microglial morphology, reveals brain region- and sex-dependent phenotypes}}, doi = {10.15479/at:ista:12378}, year = {2022}, } @article{12138, abstract = {Complex I is the first enzyme in the respiratory chain, which is responsible for energy production in mitochondria and bacteria1. Complex I couples the transfer of two electrons from NADH to quinone and the translocation of four protons across the membrane2, but the coupling mechanism remains contentious. Here we present cryo-electron microscopy structures of Escherichia coli complex I (EcCI) in different redox states, including catalytic turnover. EcCI exists mostly in the open state, in which the quinone cavity is exposed to the cytosol, allowing access for water molecules, which enable quinone movements. Unlike the mammalian paralogues3, EcCI can convert to the closed state only during turnover, showing that closed and open states are genuine turnover intermediates. The open-to-closed transition results in the tightly engulfed quinone cavity being connected to the central axis of the membrane arm, a source of substrate protons. Consistently, the proportion of the closed state increases with increasing pH. We propose a detailed but straightforward and robust mechanism comprising a ‘domino effect’ series of proton transfers and electrostatic interactions: the forward wave (‘dominoes stacking’) primes the pump, and the reverse wave (‘dominoes falling’) results in the ejection of all pumped protons from the distal subunit NuoL. This mechanism explains why protons exit exclusively from the NuoL subunit and is supported by our mutagenesis data. We contend that this is a universal coupling mechanism of complex I and related enzymes.}, author = {Kravchuk, Vladyslav and Petrova, Olga and Kampjut, Domen and Wojciechowska-Bason, Anna and Breese, Zara and Sazanov, Leonid A}, issn = {1476-4687}, journal = {Nature}, keywords = {Multidisciplinary}, number = {7928}, pages = {808--814}, publisher = {Springer Nature}, title = {{A universal coupling mechanism of respiratory complex I}}, doi = {10.1038/s41586-022-05199-7}, volume = {609}, year = {2022}, } @article{12248, abstract = {Eurasian brine shrimp (genus Artemia) have closely related sexual and asexual lineages of parthenogenetic females, which produce rare males at low frequencies. Although they are known to have ZW chromosomes, these are not well characterized, and it is unclear whether they are shared across the clade. Furthermore, the underlying genetic architecture of the transmission of asexuality, which can occur when rare males mate with closely related sexual females, is not well understood. We produced a chromosome-level assembly for the sexual Eurasian species Artemia sinica and characterized in detail the pair of sex chromosomes of this species. We combined this new assembly with short-read genomic data for the sexual species Artemia sp. Kazakhstan and several asexual lineages of Artemia parthenogenetica, allowing us to perform an in-depth characterization of sex-chromosome evolution across the genus. We identified a small differentiated region of the ZW pair that is shared by all sexual and asexual lineages, supporting the shared ancestry of the sex chromosomes. We also inferred that recombination suppression has spread to larger sections of the chromosome independently in the American and Eurasian lineages. Finally, we took advantage of a rare male, which we backcrossed to sexual females, to explore the genetic basis of asexuality. Our results suggest that parthenogenesis is likely partly controlled by a locus on the Z chromosome, highlighting the interplay between sex determination and asexuality.}, author = {Elkrewi, Marwan N and Khauratovich, Uladzislava and Toups, Melissa A and Bett, Vincent K and Mrnjavac, Andrea and Macon, Ariana and Fraisse, Christelle and Sax, Luca and Huylmans, Ann K and Hontoria, Francisco and Vicoso, Beatriz}, issn = {1943-2631}, journal = {Genetics}, keywords = {Genetics}, number = {2}, publisher = {Oxford University Press}, title = {{ZW sex-chromosome evolution and contagious parthenogenesis in Artemia brine shrimp}}, doi = {10.1093/genetics/iyac123}, volume = {222}, year = {2022}, } @inproceedings{11458, abstract = {The increasing computational requirements of deep neural networks (DNNs) have led to significant interest in obtaining DNN models that are sparse, yet accurate. Recent work has investigated the even harder case of sparse training, where the DNN weights are, for as much as possible, already sparse to reduce computational costs during training. Existing sparse training methods are often empirical and can have lower accuracy relative to the dense baseline. In this paper, we present a general approach called Alternating Compressed/DeCompressed (AC/DC) training of DNNs, demonstrate convergence for a variant of the algorithm, and show that AC/DC outperforms existing sparse training methods in accuracy at similar computational budgets; at high sparsity levels, AC/DC even outperforms existing methods that rely on accurate pre-trained dense models. An important property of AC/DC is that it allows co-training of dense and sparse models, yielding accurate sparse–dense model pairs at the end of the training process. This is useful in practice, where compressed variants may be desirable for deployment in resource-constrained settings without re-doing the entire training flow, and also provides us with insights into the accuracy gap between dense and compressed models. The code is available at: https://github.com/IST-DASLab/ACDC.}, author = {Peste, Elena-Alexandra and Iofinova, Eugenia B and Vladu, Adrian and Alistarh, Dan-Adrian}, booktitle = {35th Conference on Neural Information Processing Systems}, isbn = {9781713845393}, issn = {1049-5258}, location = {Virtual, Online}, pages = {8557--8570}, publisher = {Neural Information Processing Systems Foundation}, title = {{AC/DC: Alternating Compressed/DeCompressed training of deep neural networks}}, volume = {34}, year = {2021}, } @article{10146, abstract = {The enzymes of the mitochondrial electron transport chain are key players of cell metabolism. Despite being active when isolated, in vivo they associate into supercomplexes1, whose precise role is debated. Supercomplexes CIII2CIV1-2 (refs. 2,3), CICIII2 (ref. 4) and CICIII2CIV (respirasome)5,6,7,8,9,10 exist in mammals, but in contrast to CICIII2 and the respirasome, to date the only known eukaryotic structures of CIII2CIV1-2 come from Saccharomyces cerevisiae11,12 and plants13, which have different organization. Here we present the first, to our knowledge, structures of mammalian (mouse and ovine) CIII2CIV and its assembly intermediates, in different conformations. We describe the assembly of CIII2CIV from the CIII2 precursor to the final CIII2CIV conformation, driven by the insertion of the N terminus of the assembly factor SCAF1 (ref. 14) deep into CIII2, while its C terminus is integrated into CIV. Our structures (which include CICIII2 and the respirasome) also confirm that SCAF1 is exclusively required for the assembly of CIII2CIV and has no role in the assembly of the respirasome. We show that CIII2 is asymmetric due to the presence of only one copy of subunit 9, which straddles both monomers and prevents the attachment of a second copy of SCAF1 to CIII2, explaining the presence of one copy of CIV in CIII2CIV in mammals. Finally, we show that CIII2 and CIV gain catalytic advantage when assembled into the supercomplex and propose a role for CIII2CIV in fine tuning the efficiency of electron transfer in the electron transport chain.}, author = {Vercellino, Irene and Sazanov, Leonid A}, issn = {1476-4687}, journal = {Nature}, number = {7880}, pages = {364--367}, publisher = {Springer Nature}, title = {{Structure and assembly of the mammalian mitochondrial supercomplex CIII2CIV}}, doi = {10.1038/s41586-021-03927-z}, volume = {598}, year = {2021}, } @article{10166, abstract = {While sexual reproduction is widespread among many taxa, asexual lineages have repeatedly evolved from sexual ancestors. Despite extensive research on the evolution of sex, it is still unclear whether this switch represents a major transition requiring major molecular reorganization, and how convergent the changes involved are. In this study, we investigated the phylogenetic relationship and patterns of gene expression of sexual and asexual lineages of Eurasian Artemia brine shrimp, to assess how gene expression patterns are affected by the transition to asexuality. We find only a few genes that are consistently associated with the evolution of asexuality, suggesting that this shift may not require an extensive overhauling of the meiotic machinery. While genes with sex-biased expression have high rates of expression divergence within Eurasian Artemia, neither female- nor male-biased genes appear to show unusual evolutionary patterns after sexuality is lost, contrary to theoretical expectations.}, author = {Huylmans, Ann K and Macon, Ariana and Hontoria, Francisco and Vicoso, Beatriz}, issn = {1471-2954}, journal = {Proceedings of the Royal Society B: Biological Sciences}, keywords = {asexual reproduction, parthenogenesis, sex-biased genes, sexual conflict, automixis, crustaceans}, number = {1959}, publisher = {The Royal Society}, title = {{Transitions to asexuality and evolution of gene expression in Artemia brine shrimp}}, doi = {10.1098/rspb.2021.1720}, volume = {288}, year = {2021}, } @article{10535, abstract = {Realistic models of biological processes typically involve interacting components on multiple scales, driven by changing environment and inherent stochasticity. Such models are often analytically and numerically intractable. We revisit a dynamic maximum entropy method that combines a static maximum entropy with a quasi-stationary approximation. This allows us to reduce stochastic non-equilibrium dynamics expressed by the Fokker-Planck equation to a simpler low-dimensional deterministic dynamics, without the need to track microscopic details. Although the method has been previously applied to a few (rather complicated) applications in population genetics, our main goal here is to explain and to better understand how the method works. We demonstrate the usefulness of the method for two widely studied stochastic problems, highlighting its accuracy in capturing important macroscopic quantities even in rapidly changing non-stationary conditions. For the Ornstein-Uhlenbeck process, the method recovers the exact dynamics whilst for a stochastic island model with migration from other habitats, the approximation retains high macroscopic accuracy under a wide range of scenarios in a dynamic environment.}, author = {Bod'ová, Katarína and Szep, Eniko and Barton, Nicholas H}, issn = {1553-7358}, journal = {PLoS Computational Biology}, number = {12}, publisher = {Public Library of Science}, title = {{Dynamic maximum entropy provides accurate approximation of structured population dynamics}}, doi = {10.1371/journal.pcbi.1009661}, volume = {17}, year = {2021}, } @article{9168, abstract = {Interspecific crossing experiments have shown that sex chromosomes play a major role in reproductive isolation between many pairs of species. However, their ability to act as reproductive barriers, which hamper interspecific genetic exchange, has rarely been evaluated quantitatively compared to Autosomes. This genome-wide limitation of gene flow is essential for understanding the complete separation of species, and thus speciation. Here, we develop a mainland-island model of secondary contact between hybridizing species of an XY (or ZW) sexual system. We obtain theoretical predictions for the frequency of introgressed alleles, and the strength of the barrier to neutral gene flow for the two types of chromosomes carrying multiple interspecific barrier loci. Theoretical predictions are obtained for scenarios where introgressed alleles are rare. We show that the same analytical expressions apply for sex chromosomes and autosomes, but with different sex-averaged effective parameters. The specific features of sex chromosomes (hemizygosity and absence of recombination in the heterogametic sex) lead to reduced levels of introgression on the X (or Z) compared to autosomes. This effect can be enhanced by certain types of sex-biased forces, but it remains overall small (except when alleles causing incompatibilities are recessive). We discuss these predictions in the light of empirical data comprising model-based tests of introgression and cline surveys in various biological systems.}, author = {Fraisse, Christelle and Sachdeva, Himani}, issn = {1943-2631}, journal = {Genetics}, number = {2}, publisher = {Genetics Society of America}, title = {{The rates of introgression and barriers to genetic exchange between hybridizing species: Sex chromosomes vs autosomes}}, doi = {10.1093/genetics/iyaa025}, volume = {217}, year = {2021}, } @phdthesis{9418, abstract = {Deep learning is best known for its empirical success across a wide range of applications spanning computer vision, natural language processing and speech. Of equal significance, though perhaps less known, are its ramifications for learning theory: deep networks have been observed to perform surprisingly well in the high-capacity regime, aka the overfitting or underspecified regime. Classically, this regime on the far right of the bias-variance curve is associated with poor generalisation; however, recent experiments with deep networks challenge this view. This thesis is devoted to investigating various aspects of underspecification in deep learning. First, we argue that deep learning models are underspecified on two levels: a) any given training dataset can be fit by many different functions, and b) any given function can be expressed by many different parameter configurations. We refer to the second kind of underspecification as parameterisation redundancy and we precisely characterise its extent. Second, we characterise the implicit criteria (the inductive bias) that guide learning in the underspecified regime. Specifically, we consider a nonlinear but tractable classification setting, and show that given the choice, neural networks learn classifiers with a large margin. Third, we consider learning scenarios where the inductive bias is not by itself sufficient to deal with underspecification. We then study different ways of ‘tightening the specification’: i) In the setting of representation learning with variational autoencoders, we propose a hand- crafted regulariser based on mutual information. ii) In the setting of binary classification, we consider soft-label (real-valued) supervision. We derive a generalisation bound for linear networks supervised in this way and verify that soft labels facilitate fast learning. Finally, we explore an application of soft-label supervision to the training of multi-exit models.}, author = {Bui Thi Mai, Phuong}, issn = {2663-337X}, pages = {125}, publisher = {Institute of Science and Technology Austria}, title = {{Underspecification in deep learning}}, doi = {10.15479/AT:ISTA:9418}, year = {2021}, } @article{9431, abstract = {Inositol hexakisphosphate (IP6) is an assembly cofactor for HIV-1. We report here that IP6 is also used for assembly of Rous sarcoma virus (RSV), a retrovirus from a different genus. IP6 is ~100-fold more potent at promoting RSV mature capsid protein (CA) assembly than observed for HIV-1 and removal of IP6 in cells reduces infectivity by 100-fold. Here, visualized by cryo-electron tomography and subtomogram averaging, mature capsid-like particles show an IP6-like density in the CA hexamer, coordinated by rings of six lysines and six arginines. Phosphate and IP6 have opposing effects on CA in vitro assembly, inducing formation of T = 1 icosahedrons and tubes, respectively, implying that phosphate promotes pentamer and IP6 hexamer formation. Subtomogram averaging and classification optimized for analysis of pleomorphic retrovirus particles reveal that the heterogeneity of mature RSV CA polyhedrons results from an unexpected, intrinsic CA hexamer flexibility. In contrast, the CA pentamer forms rigid units organizing the local architecture. These different features of hexamers and pentamers determine the structural mechanism to form CA polyhedrons of variable shape in mature RSV particles.}, author = {Obr, Martin and Ricana, Clifton L. and Nikulin, Nadia and Feathers, Jon-Philip R. and Klanschnig, Marco and Thader, Andreas and Johnson, Marc C. and Vogt, Volker M. and Schur, Florian KM and Dick, Robert A.}, issn = {2041-1723}, journal = {Nature Communications}, keywords = {General Biochemistry, Genetics and Molecular Biology, General Physics and Astronomy, General Chemistry}, number = {1}, publisher = {Nature Research}, title = {{Structure of the mature Rous sarcoma virus lattice reveals a role for IP6 in the formation of the capsid hexamer}}, doi = {10.1038/s41467-021-23506-0}, volume = {12}, year = {2021}, }