Showing papers by "Moscow Institute of Physics and Technology published in 2018"

CellProfiler 3.0: Next-generation image processing for biology.

Abstract: CellProfiler has enabled the scientific research community to create flexible, modular image analysis pipelines since its release in 2005. Here, we describe CellProfiler 3.0, a new version of the software supporting both whole-volume and plane-wise analysis of three-dimensional (3D) image stacks, increasingly common in biomedical research. CellProfiler's infrastructure is greatly improved, and we provide a protocol for cloud-based, large-scale image processing. New plugins enable running pretrained deep learning models on images. Designed by and for biologists, CellProfiler equips researchers with powerful computational tools via a well-documented user interface, empowering biologists in all fields to create quantitative, reproducible image analysis workflows.

Analysis of shared heritability in common disorders of the brain

Abstract: Disorders of the brain can exhibit considerable epidemiological comorbidity and often share symptoms, provoking debate about their etiologic overlap. We quantified the genetic sharing of 25 brain disorders from genome-wide association studies of 265,218 patients and 784,643 control participants and assessed their relationship to 17 phenotypes from 1,191,588 individuals. Psychiatric disorders share common variant risk, whereas neurological disorders appear more distinct from one another and from the psychiatric disorders. We also identified significant sharing between disorders and a number of brain phenotypes, including cognitive measures. Further, we conducted simulations to explore how statistical power, diagnostic misclassification, and phenotypic heterogeneity affect genetic correlations. These results highlight the importance of common genetic variation as a risk factor for brain disorders and the value of heritability-based methods in understanding their etiology.

Deep reinforcement learning for de novo drug design

Abstract: We have devised and implemented a novel computational strategy for de novo design of molecules with desired properties termed ReLeaSE (Reinforcement Learning for Structural Evolution). On the basis of deep and reinforcement learning (RL) approaches, ReLeaSE integrates two deep neural networks—generative and predictive—that are trained separately but are used jointly to generate novel targeted chemical libraries. ReLeaSE uses simple representation of molecules by their simplified molecular-input line-entry system (SMILES) strings only. Generative models are trained with a stack-augmented memory network to produce chemically feasible SMILES strings, and predictive models are derived to forecast the desired properties of the de novo–generated compounds. In the first phase of the method, generative and predictive models are trained separately with a supervised learning algorithm. In the second phase, both models are trained jointly with the RL approach to bias the generation of new chemical structures toward those with the desired physical and/or biological properties. In the proof-of-concept study, we have used the ReLeaSE method to design chemical libraries with a bias toward structural complexity or toward compounds with maximal, minimal, or specific range of physical properties, such as melting point or hydrophobicity, or toward compounds with inhibitory activity against Janus protein kinase 2. The approach proposed herein can find a general use for generating targeted chemical libraries of novel compounds optimized for either a single desired property or multiple properties.

HOCOMOCO: towards a complete collection of transcription factor binding models for human and mouse via large-scale ChIP-Seq analysis

Abstract: We present a major update of the HOCOMOCO collection that consists of patterns describing DNA binding specificities for human and mouse transcription factors. In this release, we profited from a nearly doubled volume of published in vivo experiments on transcription factor (TF) binding to expand the repertoire of binding models, replace low-quality models previously based on in vitro data only and cover more than a hundred TFs with previously unknown binding specificities. This was achieved by systematic motif discovery from more than five thousand ChIP-Seq experiments uniformly processed within the BioUML framework with several ChIP-Seq peak calling tools and aggregated in the GTRD database. HOCOMOCO v11 contains binding models for 453 mouse and 680 human transcription factors and includes 1302 mononucleotide and 576 dinucleotide position weight matrices, which describe primary binding preferences of each transcription factor and reliable alternative binding specificities. An interactive interface and bulk downloads are available on the web: http://hocomoco.autosome.ru and http://www.cbrc.kaust.edu.sa/hocomoco11. In this release, we complement HOCOMOCO by MoLoTool (Motif Location Toolbox, http://molotool.autosome.ru) that applies HOCOMOCO models for visualization of binding sites in short DNA sequences.

Identification of heavy-flavour jets with the CMS detector in pp collisions at 13 TeV

Abstract: Many measurements and searches for physics beyond the standard model at the LHC rely on the efficient identification of heavy-flavour jets, i.e. jets originating from bottom or charm quarks. In this paper, the discriminating variables and the algorithms used for heavy-flavour jet identification during the first years of operation of the CMS experiment in proton-proton collisions at a centre-of-mass energy of 13 TeV, are presented. Heavy-flavour jet identification algorithms have been improved compared to those used previously at centre-of-mass energies of 7 and 8 TeV. For jets with transverse momenta in the range expected in simulated events, these new developments result in an efficiency of 68% for the correct identification of a b jet for a probability of 1% of misidentifying a light-flavour jet. The improvement in relative efficiency at this misidentification probability is about 15%, compared to previous CMS algorithms. In addition, for the first time algorithms have been developed to identify jets containing two b hadrons in Lorentz-boosted event topologies, as well as to tag c jets. The large data sample recorded in 2016 at a centre-of-mass energy of 13 TeV has also allowed the development of new methods to measure the efficiency and misidentification probability of heavy-flavour jet identification algorithms. The b jet identification efficiency is measured with a precision of a few per cent at moderate jet transverse momenta (between 30 and 300 GeV) and about 5% at the highest jet transverse momenta (between 500 and 1000 GeV).

CatBoost: unbiased boosting with categorical features

Abstract: This paper presents the key algorithmic techniques behind CatBoost, a new gradient boosting toolkit. Their combination leads to CatBoost outperforming other publicly available boosting implementations in terms of quality on a variety of datasets. Two critical algorithmic advances introduced in CatBoost are the implementation of ordered boosting, a permutation-driven alternative to the classic algorithm, and an innovative algorithm for processing categorical features. Both techniques were created to fight a prediction shift caused by a special kind of target leakage present in all currently existing implementations of gradient boosting algorithms. In this paper, we provide a detailed analysis of this problem and demonstrate that proposed algorithms solve it effectively, leading to excellent empirical results.

Novel Nanostructures and Materials for Strong Light–Matter Interactions

Abstract: Quantum mechanical interactions between electromagnetic radiation and matter underlie a broad spectrum of optical phenomena. Strong light-matter interactions result in the well-known vacuum Rabi splitting and emergence of new polaritonic eigenmodes of the coupled system. Thanks to recent progress in nanofabrication, observation of strong coupling has become possible in a great variety of optical nanostructures. Here, we review recently studied and emerging materials for realization of strong light–matter interactions. We present general theoretical formalism describing strong coupling and give an overview of various photonic structures and materials allowing for realization of this regime, including plasmonic and dielectric nanoantennas, novel two-dimensional materials, carbon nanotubes, and molecular vibrational transitions. In addition, we discuss practical applications that can benefit from these effects and give an outlook on unsettled questions that remain open for future research.

Search for dark matter and other new phenomena in events with an energetic jet and large missing transverse momentum using the ATLAS detector

Abstract: Results of a search for new phenomena in final states with an energetic jet and large missing transverse momentum are reported. The search uses proton-proton collision data corresponding to an integrated luminosity of 36.1 fb−1 at a centre-of-mass energy of 13 TeV collected in 2015 and 2016 with the ATLAS detector at the Large Hadron Collider. Events are required to have at least one jet with a transverse momentum above 250 GeV and no leptons (e or μ). Several signal regions are considered with increasing requirements on the missing transverse momentum above 250 GeV. Good agreement is observed between the number of events in data and Standard Model predictions. The results are translated into exclusion limits in models with pair-produced weakly interacting dark-matter candidates, large extra spatial dimensions, and supersymmetric particles in several compressed scenarios.

Performance of the CMS muon detector and muon reconstruction with proton-proton collisions at s=13 TeV

Abstract: The CMS muon detector system, muon reconstruction software, and high-level trigger underwent significant changes in 2013–2014 in preparation for running at higher LHC collision energy and instantaneous luminosity. The performance of the modified system is studied using proton-proton collision data at center-of-mass energy √s=13 TeV, collected at the LHC in 2015 and 2016. The measured performance parameters, including spatial resolution, efficiency, and timing, are found to meet all design specifications and are well reproduced by simulation. Despite the more challenging running conditions, the modified muon system is found to perform as well as, and in many aspects better than, previously. We dedicate this paper to the memory of Prof. Alberto Benvenuti, whose work was fundamental for the CMS muon detector.

Reinforced Adversarial Neural Computer for de Novo Molecular Design

Abstract: In silico modeling is a crucial milestone in modern drug design and development. Although computer-aided approaches in this field are well-studied, the application of deep learning methods in this research area is at the beginning. In this work, we present an original deep neural network (DNN) architecture named RANC (Reinforced Adversarial Neural Computer) for the de novo design of novel small-molecule organic structures based on the generative adversarial network (GAN) paradigm and reinforcement learning (RL). As a generator RANC uses a differentiable neural computer (DNC), a category of neural networks, with increased generation capabilities due to the addition of an explicit memory bank, which can mitigate common problems found in adversarial settings. The comparative results have shown that RANC trained on the SMILES string representation of the molecules outperforms its first DNN-based counterpart ORGANIC by several metrics relevant to drug discovery: the number of unique structures, passing medicin...

Search for Heavy Neutral Leptons in Events with Three Charged Leptons in Proton-Proton Collisions at √s = 13 TeV

Abstract: A search for a heavy neutral lepton N of Majorana nature decaying into a W boson and a charged lepton is performed using the CMS detector at the LHC. The targeted signature consists of three prompt charged leptons in any flavor combination of electrons and muons. The data were collected in proton-proton collisions at a center-of-mass energy of 13 TeV, with an integrated luminosity of 35.9 fb^(−1). The search is performed in the N mass range between 1 GeV and 1.2 TeV. The data are found to be consistent with the expected standard model background. Upper limits are set on the values of |V_(eN)|^2and |V_(μN)|^2, where V_(lN) is the matrix element describing the mixing of N with the standard model neutrino of flavor l. These are the first direct limits for N masses above 500 GeV and the first limits obtained at a hadron collider for N masses below 40 GeV.

Gravitational Bose-Einstein condensation in the kinetic regime

Abstract: We study Bose-Einstein condensation and the formation of Bose stars in virialized dark matter halos and miniclusters by universal gravitational interactions. We prove that this phenomenon does occur and it is described by a kinetic equation. We give an expression for the condensation time. Our results suggest that Bose stars may form kinetically in mainstream dark matter models such as invisible QCD axions and fuzzy dark matter.

Search for Higgs boson pair production in the γ γ bb¯ final state with 13 TeV pp collision data collected by the ATLAS experiment

Abstract: A search is performed for resonant and non-resonant Higgs boson pair production in the $ \upgamma \upgamma b\overline{b} $ final state. The data set used corresponds to an integrated luminosity of 36.1 fb$^{−1}$ of proton-proton collisions at a centre-of-mass energy of 13 TeV recorded by the ATLAS detector at the CERN Large Hadron Collider. No significant excess relative to the Standard Model expectation is observed. The observed limit on the non-resonant Higgs boson pair cross-section is 0.73 pb at 95% confidence level. This observed limit is equivalent to 22 times the predicted Standard Model cross-section. The Higgs boson self-coupling (κ$_{λ}$ = λ$_{HHH}$/λ$_{HHH}^{SM}$ ) is constrained at 95% confidence level to −8.2 < κ$_{λ}$ < 13.2. For resonant Higgs boson pair production through $ X\to HH\to \upgamma \upgamma b\overline{b} $ , the limit is presented, using the narrow-width approximation, as a function of m$_{X}$ in the range 260 GeV < m$_{X}$ < 1000 GeV. The observed limits range from 1.1 pb to 0.12 pb over this mass range.

Time-domain effective-one-body gravitational waveforms for coalescing compact binaries with nonprecessing spins, tides, and self-spin effects

Abstract: We present TEOBResumS, a new effective-one-body (EOB) waveform model for nonprecessing (spin-aligned) and tidally interacting compact binaries. Spin-orbit and spin-spin effects are blended together by making use of the concept of centrifugal EOB radius. The point-mass sector through merger and ringdown is informed by numerical relativity (NR) simulations of binary black holes (BBHs) computed with the SpEC and bam codes. An improved, NR-based phenomenological description of the postmerger waveform is developed. The tidal sector of TEOBResumS describes the dynamics of neutron star binaries up to merger and incorporates a resummed attractive potential motivated by recent advances in the post-Newtonian and gravitational self-force description of relativistic tidal interactions. Equation-of-state-dependent self-spin interactions (monopole-quadrupole effects) are incorporated in the model using leading order post-Newtonian results in a new expression of the centrifugal radius. TEOBResumS is compared to 135 SpEC and 19 bam BBH waveforms. The maximum unfaithfulness to SpEC data ¯ F—at design Advanced LIGO sensitivity and evaluated with total mass M with a variance of 10M⊙≤M≤200M⊙—is always below 2.5×10−3 except for a single outlier that grazes the 7.1×10−3 level. When compared to bam data, ¯ F is smaller than 0.01 except for a single outlier in one of the corners of the NR-covered parameter space that reaches the 0.052 level. TEOBResumS is also compatible, up to merger, to high-end NR waveforms from binary neutron stars with spin effects and reduced initial eccentricity computed with the bam and thc codes. The data quality of binary neutron star waveforms is assessed via rigorous convergence tests from multiple resolution runs and takes into account systematic effects estimated by using the two independent high-order NR codes. The model is designed to generate accurate templates for the analysis of LIGO-Virgo data through merger and ringdown. We demonstrate its use by analyzing the publicly available data for GW150914.

Improved Ferroelectric Switching Endurance of La-Doped Hf0.5Zr0.5O2 Thin Films

Abstract: Hf0.5Zr0.5O2 thin films are one of the most appealing HfO2-based ferroelectric thin films, which have been researched extensively for their applications in ferroelectric memory devices. In this work, a 1 mol % La-doped Hf0.5Zr0.5O2 thin film was grown by plasma-assisted atomic layer deposition and annealed at temperatures of 450 and 500 °C to crystallize the film into the desired orthorhombic phase. Despite the use of a lower temperature than that used in previous reports, the film showed highly promising ferroelectric properties—a remnant polarization of ∼30 μC/cm2 and switching cycle endurance up to 4 × 1010. The performance was much better than that of undoped Hf0.5Zr0.5O2 thin films, demonstrating the positive influence of La doping. Such improvements were mainly attributed to the decreased coercive field (by ∼30% compared to the undoped film), which allowed for the use of a lower applied field to drive the cycling tests while maintaining a high polarization value. La doping also decreased the leakage...

Entangled Conditional Adversarial Autoencoder for de Novo Drug Discovery

Abstract: Modern computational approaches and machine learning techniques accelerate the invention of new drugs. Generative models can discover novel molecular structures within hours, while conventional drug discovery pipelines require months of work. In this article, we propose a new generative architecture, entangled conditional adversarial autoencoder, that generates molecular structures based on various properties, such as activity against a specific protein, solubility, or ease of synthesis. We apply the proposed model to generate a novel inhibitor of Janus kinase 3, implicated in rheumatoid arthritis, psoriasis, and vitiligo. The discovered molecule was tested in vitro and showed good activity and selectivity.

Search for narrow and broad dijet resonances in proton-proton collisions at √s=13 TeV and constraints on dark matter mediators and other new particles

Abstract: Searches for resonances decaying into pairs of jets are performed using proton-proton collision data collected at s√=13 TeV corresponding to an integrated luminosity of up to 36 fb−1. A low-mass search, for resonances with masses between 0.6 and 1.6 TeV, is performed based on events with dijets reconstructed at the trigger level from calorimeter information. A high-mass search, for resonances with masses above 1.6 TeV, is performed using dijets reconstructed offline with a particle-flow algorithm. The dijet mass spectrum is well described by a smooth parameterization and no evidence for the production of new particles is observed. Upper limits at 95% confidence level are reported on the production cross section for narrow resonances with masses above 0.6 TeV. In the context of specific models, the limits exclude string resonances with masses below 7.7 TeV, scalar diquarks below 7.2 TeV, axigluons and colorons below 6.1 TeV, excited quarks below 6.0 TeV, color-octet scalars below 3.4 TeV, W′ bosons below 3.3 TeV, Z′ bosons below 2.7 TeV, Randall-Sundrum gravitons below 1.8 TeV and in the range 1.9 to 2.5 TeV, and dark matter mediators below 2.6 TeV. The limits on both vector and axial-vector mediators, in a simplified model of interactions between quarks and dark matter particles, are presented as functions of dark matter particle mass and coupling to quarks. Searches are also presented for broad resonances, including for the first time spin-1 resonances with intrinsic widths as large as 30% of the resonance mass. The broad resonance search improves and extends the exclusions of a dark matter mediator to larger values of its mass and coupling to quarks.

High-harmonic spectroscopy of ultrafast many-body dynamics in strongly correlated systems

Abstract: We bring together two topics that, until now, have been the focus of intense but non-overlapping research efforts. The first concerns high-harmonic generation in solids, which occurs when an intense light field excites a highly non-equilibrium electronic response in a semiconductor or a dielectric. The second concerns many-body dynamics in strongly correlated systems such as the Mott insulator. We show that high-harmonic generation can be used to time-resolve ultrafast many-body dynamics associated with an optically driven phase transition, with accuracy far exceeding one cycle of the driving light field. Our work paves the way for time-resolving highly non-equilibrium many-body dynamics in strongly correlated systems, with few femtosecond accuracy. A highly nonlinear optical response can be used to time-resolve light-induced phase transitions with few-femtosecond to sub-femtosecond accuracy, paving the way for time-resolving highly correlated many-body dynamics in strongly correlated systems with few-femtosecond accuracy.

Resonant Terahertz Detection Using Graphene Plasmons

Abstract: Plasmons, collective oscillations of electron systems, can efficiently couple light and electric current, and thus can be used to create sub-wavelength photodetectors, radiation mixers, and on-chip spectrometers. Despite considerable effort, it has proven challenging to implement plasmonic devices operating at terahertz frequencies. The material capable to meet this challenge is graphene as it supports long-lived electrically-tunable plasmons. Here we demonstrate plasmon-assisted resonant detection of terahertz radiation by antenna-coupled graphene transistors that act as both plasmonic Fabry-Perot cavities and rectifying elements. By varying the plasmon velocity using gate voltage, we tune our detectors between multiple resonant modes and exploit this functionality to measure plasmon wavelength and lifetime in bilayer graphene as well as to probe collective modes in its moire minibands. Our devices offer a convenient tool for further plasmonic research that is often exceedingly difficult under non-ambient conditions (e.g. cryogenic temperatures and strong magnetic fields) and promise a viable route for various photonic applications.

Combined search for electroweak production of charginos and neutralinos in proton-proton collisions at √s=13 TeV

Abstract: A statistical combination of several searches for the electroweak production of charginos and neutralinos is presented. All searches use proton-proton collision data at $ \sqrt{s}=13 $ TeV, recorded with the CMS detector at the LHC in 2016 and corresponding to an integrated luminosity of 35.9 fb$^{−1}$. In addition to the combination of previous searches, a targeted analysis requiring three or more charged leptons (electrons or muons) is presented, focusing on the challenging scenario in which the difference in mass between the two least massive neutralinos is approximately equal to the mass of the Z boson. The results are interpreted in simplified models of chargino-neutralino or neutralino pair production. For chargino-neutralino production, in the case when the lightest neutralino is massless, the combination yields an observed (expected) limit at the 95% confidence level on the chargino mass of up to 650 (570) GeV, improving upon the individual analysis limits by up to 40 GeV. If the mass difference between the two least massive neutralinos is approximately equal to the mass of the Z boson in the chargino-neutralino model, the targeted search requiring three or more leptons obtains observed and expected exclusion limits of around 225 GeV on the second neutralino mass and 125 GeV on the lightest neutralino mass, improving the observed limit by about 60 GeV in both masses compared to the previous CMS result. In the neutralino pair production model, the combined observed (expected) exclusion limit on the neutralino mass extends up to 650–750 (550–750) GeV, depending on the branching fraction assumed. This extends the observed exclusion achieved in the individual analyses by up to 200 GeV. The combined result additionally excludes some intermediate gaps in the mass coverage of the individual analyses.

Narrow-linewidth lasing and soliton Kerr microcombs with ordinary laser diodes

Abstract: Narrow-linewidth lasers and optical frequency combs generated with mode-locked lasers have revolutionized optical frequency metrology. The advent of soliton Kerr frequency combs in compact crystalline or integrated ring optical microresonators has opened new horizons in academic research and industrial applications. These combs, as was naturally assumed, however, require narrow-linewidth, single-frequency pump lasers. We demonstrate that an ordinary cost-effective broadband Fabry–Perot laser diode at the hundreds of milliwatts level, self-injection-locked to a microresonator, can be efficiently transformed to a powerful single-frequency, ultra-narrow-linewidth light source with further transformation to a coherent soliton comb oscillator. Our findings pave the way to the most compact and inexpensive highly coherent lasers, frequency comb sources, and comb-based devices for mass production. A broadband multi-frequency Fabry–Perot laser diode, when coupled to a high-Q microresonator, can be efficiently transformed to an ~100 mW narrow-linewidth single-frequency light source, and subsequently, to a coherent soliton Kerr comb oscillator.