Search

Recent advancements in specialized large-scale architectures for training image and language have profoundly impacted the field of computer vision and natural language processing (NLP). Language models, such as the recent ChatGPT and GPT4 have demonstrated exceptional capabilities in processing, translating, and generating human languages. These breakthroughs have also been reflected in protein research, leading to the rapid development of numerous new methods in a short time, with unprecedented performance. Language models, in particular, have seen widespread use in protein research, as they have been utilized to embed proteins, generate novel ones, and predict tertiary structures. In this book chapter, we provide an overview of the use of protein generative models, reviewing 1) language models for the design of novel artificial proteins, 2) works that use non-Transformer architectures, and 3) applications in directed evolution approaches., Comment: 16 pages, 4 figures, 2 tables

Vergangene Naturkatastrophen zeigen, dass Katastrophenbewältigung ohne Berücksichtigung der Zivilbevölkerung nicht mehr denkbar ist. Sogenannte Spontanhelfende organisieren und koordinieren sich und ihre Hilfe dabei weitestgehend selbstständig. Ihr Zustrom an Einsatzorten hat in der Vergangenheit jedoch zu massiven Überlastungen von Einsatzorten geführt, wodurch offizielles Katastrophenpersonal vor Ort in der Bewältigung ihrer Arbeit durch diese zusätzliche Herausforderung behindert wurde. Die Probleme, die durch den Zustrom an Einsatzorten entstanden sind, waren vor allem auf die unvorhersehbare Natur der Spontanhelfenden und das Fehlen von Katastrophenmanagementwerkzeugen zur Planung und Vorbereitung solcher Szenarien zurückzuführen. Unter Verwendung eines designwissenschaftlichen Forschungsansatzes zielt die Dissertation darauf ab, ein Informationssystem zu entwerfen, zu entwickeln und zu evaluieren, das den Zustrom von Spontanhelfenden an Einsatzorten vorhersagen soll., ePast natural disasters showed that disaster management is no longer conceivable without considering civil participation, i.e., spontaneous volunteers, who predominantly organize and coordinate their help efforts on their own. In this regard, spontaneous volunteers led to, e.g., massive overloads of operating sites, causing official on-site disaster staff to be hampered in their work by this additional challenge. The problems caused by their influx at operating sites were primarily due to the unpredictable nature of spontaneous volunteers and a lack of disaster management tools to plan and train for such scenarios. Following a design science research approach, the dissertation aims to design, develop, and evaluate an information system for predicting the influx of spontaneous volunteers at operating sites. engl. Zsfassung

The OMNeT++ simulator is well-suited for the simulation of randomized user behavior in communication networks. However, there are scenarios, where such a random model is unsuited to evaluate a communication system, and this paper attempts to highlight such a case. Using this example of ad-hoc communication between aircraft mid-flight, a tutorial-style description is attempted that shall show how the OMNeT++ simulator can be used when a wealth of real-world trace data is available. In particular, it is described how mobility trace files can be directly used within OMNeT++, and how to link the generation of data messages to this mobility data. This is explained via an example simulation that evaluates a communication network in which an aircraft notifies the ground control when it enters or leaves a specific geographic region. Additionally, a novel trace-based application has been developed to achieve this link between mobility and message generation. Furthermore, a new TDMA-based medium access protocol for decentralized communication networks is presented, which is oracle-based and thus allows a TDMA-like behavior of medium access without causing any overhead; it can be useful when upper-layer protocols should be evaluated under the assumption of TDMA-like behavior, but isolated from the effects of a full-fledged TDMA protocol. Finally, physical layer behavior is often either overly simplistic or overly computationally expensive. For the latter case, when a detailed channel model is available but its evaluation requires prohibitive computational effort, then averaging its behavior into trace data can find a middle ground between efficient evaluation and realistic representation. Hence, a novel trace-based radio model has been developed that makes use of an SNR to PER mapping. In the spirit of open science, all implementations have been made available under open licenses., Published in: M. Marek, G. Nardini, V. Vesely (Eds.), Proceedings of the 8th OMNeT++ Community Summit, Virtual Summit, September 8-10, 2021

A new international facility for antiproton and ion research (FAIR) is currently under construction in Darmstadt, Germany. The high intensity primary beam required for different research experiments will be provided by the SIS100 heavy ion synchrotron. The synchrotron is composed of fast cycling superconducting magnets from which about 300 will be integrated in Quadrupole Doublet Modules (QDM). Each module consists of two units composed of a quadrupole and corrector magnets. The First of Series Quadrupole Doublet Module was delivered to the test facility at GSI in November 2019. The assembled doublet was subjected to a dedicated test program to verify the functionality of the module components at cryogenic temperature and operating conditions. The results obtained during the testing campaign will be presented., Proceedings of the 12th International Particle Accelerator Conference, IPAC2021, Campinas, SP, Brazil

The divisor class group of a hyperelliptic curve over a field is a finite abelian group at the center of many open questions in algebraic geometry and number theory. Sutherland surveys some of these, including the computation of the associated L-functions and zeta functions used in his investigation of Sato-Tate distributions. Many of these problems lend themselves to empirical investigation, and as emphasized by Sutherland, fast arithmetic in the divisor class group is crucial for their efficiency. Indeed, implementations of these fundamental operations are at the core of the algebraic geometry packages of widely-used computer algebra systems such as Magma and Sage. This thesis provides contributions to improve the efficiency of divisor class arithmetic on hyperelliptic curves, with special attention to the often ignored split model cases and to genus 2 and 3. There are two main contributions: the introduction of “Balanced NUCOMP” for improved arithmetic on curves given by split models of arbitrary genus, and improved explicit formulas for genus 2 (ramified and split models) and genus 3 (split models) based on Balanced NUCOMP for split models and NUCOMP for ramified models. Empirical analysis, using a complete Magma implementation and testing suite, is conducted with all contributed algorithms to provide proof of correctness and comparisons to previous best. Our results show that Balanced NUCOMP does offer improvements to split model arithmetic, further narrowing the performance gap between split and ramified models. Our explicit formulas require fewer field operations, most notably significantly fewer additions, than previous best for computing arithmetic on genus 2 (ramified and split models), and genus 3 (split models). Our empirical results demonstrate that our formulas yield the fastest general-purpose arithmetic for these cases.

Cognitive Radios (CRs) tackle the spectrum scarcity problem by allowing unlicensed access on already-licensed spectrum. Where primary user (PU) radio systems hold a privileged medium access, secondary users (SUs) try to utilize and use spare time-frequency resources to establish communication. At their core, CRs need to sense, detect and predict the medium access pattern of PUs in order to facilitate communication. A promising approach for inferring these predictions is the use of Machine Learning techniques and in particular Artifical Neural Networks (ANNs). ANNs try to learn a mapping from a set of inputs to a specific desired output and can therefore be directly applied to the problem of PU activity prediction. Especially Recurrent Neural Networks (RNNs) show adequate performance for noisy measurement data and prolonged training periods. In this work, the applicability of ANN-based channel state prediction is examined via a case study on the upcoming aeronautical communication technology L-band Digital Aeronautical Communications System (LDACS). LDACS is envisioned to reuse spectral resources currently primarily allocated to other aeronautical systems such as the Distance Measuring Equipment (DME) system. Primary and secondary users are modeled with respect to preliminary LDACS specifications and the performance of the proposed algorithms is evaluated via simulation. We show that RNNs can function as prediction agents for SU LDACS medium access and that stringent reliability and interference requirements can be met. A supervised learning problem, together with an incremental learning strategy is proposed to address time-varying PU channel access patterns. Finally, a brief discussion on the use of predictions in a distributed scheduling approach is given.

The Limited Relative Error algorithm is an alternative statistical method for data evaluation. Through online result analysis it continuously requests more samples until it deems the evaluation confident enough. With this it allows researchers to hand over the control of simulation time to the algorithm, and through a-priori configuration the target result resolution is set so that arbitrarily rare events can be investigated. We provide a new description of the method as well as a stand-alone implementation and an integration of the algorithm into the OMNeT++ simulator.

In this article, we present a user-friendly web interface for two alignment-free sequence-comparison methods that we recently developed. Most alignment-free methods rely on exact word matches to estimate pairwise similarities or distances between the input sequences. By contrast, our new algorithms are based on inexact word matches. The first of these approaches uses the relative frequencies of so-called spaced words in the input sequences, i.e. words containing 'don't care' or 'wildcard' symbols at certain pre-defined positions. Various distance measures can then be defined on sequences based on their different spaced-word composition. Our second approach defines the distance between two sequences by estimating for each position in the first sequence the length of the longest substring at this position that also occurs in the second sequence with up to k mismatches. Both approaches take a set of deoxyribonucleic acid (DNA) or protein sequences as input and return a matrix of pairwise distance values that can be used as a starting point for clustering algorithms or distance-based phylogeny reconstruction. The two alignment-free programmes are accessible through a web interface at 'Göttingen Bioinformatics Compute Server (GOBICS)': http://spaced.gobics.de http://kmacs.gobics.de and the source codes can be downloaded. Open-Access-Publikationsfonds 2014 peerReviewed

In this article, we present a user-friendly web interface for two alignment-free sequence-comparison methods that we recently developed. Most alignment-free methods rely on exact word matches to estimate pairwise similarities or distances between the input sequences. By contrast, our new algorithms are based on inexact word matches. The first of these approaches uses the relative frequencies of so-called spaced words in the input sequences, i.e. words containing 'don't care' or 'wildcard' symbols at certain pre-defined positions. Various distance measures can then be defined on sequences based on their different spaced-word composition. Our second approach defines the distance between two sequences by estimating for each position in the first sequence the length of the longest substring at this position that also occurs in the second sequence with up to k mismatches. Both approaches take a set of deoxyribonucleic acid (DNA) or protein sequences as input and return a matrix of pairwise distance values that can be used as a starting point for clustering algorithms or distance-based phylogeny reconstruction.

Alignment-free methods are increasingly used for genome analysis and phylogeny reconstruction since they circumvent various difficulties of traditional approaches that rely on multiple sequence alignments. In particular, they are much faster than alignment-based methods. Most alignment-free approaches work by analyzing the k-mer composition of sequences. In this paper, we propose to use 'spaced k-mers', i.e. patterns of deterministic and 'don't care' positions instead of contiguous k-mers. Using simulated and real-world sequence data, we demonstrate that this approach produces better phylogenetic trees than alignment-free methods that rely on contiguous k-mers. In addition, distances calculated with spaced k-mers appear to be statistically more stable than distances based on contiguous k-mers.