@article{AltenburgGieseWangetal.2022, author = {Altenburg, Tom and Giese, Sven Hans-Joachim and Wang, Shengbo and Muth, Thilo and Renard, Bernhard Y.}, title = {Ad hoc learning of peptide fragmentation from mass spectra enables an interpretable detection of phosphorylated and cross-linked peptides}, series = {Nature machine intelligence}, volume = {4}, journal = {Nature machine intelligence}, number = {4}, publisher = {Springer Nature Publishing}, address = {London}, issn = {2522-5839}, doi = {10.1038/s42256-022-00467-7}, pages = {378 -- 388}, year = {2022}, abstract = {Fragmentation of peptides leaves characteristic patterns in mass spectrometry data, which can be used to identify protein sequences, but this method is challenging for mutated or modified sequences for which limited information exist. Altenburg et al. use an ad hoc learning approach to learn relevant patterns directly from unannotated fragmentation spectra. Mass spectrometry-based proteomics provides a holistic snapshot of the entire protein set of living cells on a molecular level. Currently, only a few deep learning approaches exist that involve peptide fragmentation spectra, which represent partial sequence information of proteins. Commonly, these approaches lack the ability to characterize less studied or even unknown patterns in spectra because of their use of explicit domain knowledge. Here, to elevate unrestricted learning from spectra, we introduce 'ad hoc learning of fragmentation' (AHLF), a deep learning model that is end-to-end trained on 19.2 million spectra from several phosphoproteomic datasets. AHLF is interpretable, and we show that peak-level feature importance values and pairwise interactions between peaks are in line with corresponding peptide fragments. We demonstrate our approach by detecting post-translational modifications, specifically protein phosphorylation based on only the fragmentation spectrum without a database search. AHLF increases the area under the receiver operating characteristic curve (AUC) by an average of 9.4\% on recent phosphoproteomic data compared with the current state of the art on this task. Furthermore, use of AHLF in rescoring search results increases the number of phosphopeptide identifications by a margin of up to 15.1\% at a constant false discovery rate. To show the broad applicability of AHLF, we use transfer learning to also detect cross-linked peptides, as used in protein structure analysis, with an AUC of up to 94\%.}, language = {en} } @article{CrawfordKaramatLehotaietal.2020, author = {Crawford, Tim and Karamat, Fazeelat and Lehotai, N{\´o}ra and Rentoft, Matilda and Blomberg, Jeanette and Strand, {\AA}sa and Bj{\"o}rklund, Stefan}, title = {Specific functions for mediator complex subunits from different modules in the transcriptional response of arabidopsis thaliana to abiotic stress}, series = {Scientific reports}, volume = {10}, journal = {Scientific reports}, number = {1}, publisher = {Macmillan Publishers Limited, part of Springer Nature}, address = {London}, issn = {2045-2322}, doi = {10.1038/s41598-020-61758-w}, pages = {1 -- 18}, year = {2020}, abstract = {Adverse environmental conditions are detrimental to plant growth and development. Acclimation to abiotic stress conditions involves activation of signaling pathways which often results in changes in gene expression via networks of transcription factors (TFs). Mediator is a highly conserved co-regulator complex and an essential component of the transcriptional machinery in eukaryotes. Some Mediator subunits have been implicated in stress-responsive signaling pathways; however, much remains unknown regarding the role of plant Mediator in abiotic stress responses. Here, we use RNA-seq to analyze the transcriptional response of Arabidopsis thaliana to heat, cold and salt stress conditions. We identify a set of common abiotic stress regulons and describe the sequential and combinatorial nature of TFs involved in their transcriptional regulation. Furthermore, we identify stress-specific roles for the Mediator subunits MED9, MED16, MED18 and CDK8, and putative TFs connecting them to different stress signaling pathways. Our data also indicate different modes of action for subunits or modules of Mediator at the same gene loci, including a co-repressor function for MED16 prior to stress. These results illuminate a poorly understood but important player in the transcriptional response of plants to abiotic stress and identify target genes and mechanisms as a prelude to further biochemical characterization.}, language = {en} } @article{TrautmannZhouBrahmsetal.2021, author = {Trautmann, Justin and Zhou, Lin and Brahms, Clemens Markus and Tunca, Can and Ersoy, Cem and Granacher, Urs and Arnrich, Bert}, title = {TRIPOD}, series = {Data : open access ʻData in scienceʼ journal}, volume = {6}, journal = {Data : open access ʻData in scienceʼ journal}, number = {9}, publisher = {MDPI}, address = {Basel}, issn = {2306-5729}, doi = {10.3390/data6090095}, pages = {19}, year = {2021}, abstract = {Inertial measurement units (IMUs) enable easy to operate and low-cost data recording for gait analysis. When combined with treadmill walking, a large number of steps can be collected in a controlled environment without the need of a dedicated gait analysis laboratory. In order to evaluate existing and novel IMU-based gait analysis algorithms for treadmill walking, a reference dataset that includes IMU data as well as reliable ground truth measurements for multiple participants and walking speeds is needed. This article provides a reference dataset consisting of 15 healthy young adults who walked on a treadmill at three different speeds. Data were acquired using seven IMUs placed on the lower body, two different reference systems (Zebris FDMT-HQ and OptoGait), and two RGB cameras. Additionally, in order to validate an existing IMU-based gait analysis algorithm using the dataset, an adaptable modular data analysis pipeline was built. Our results show agreement between the pressure-sensitive Zebris and the photoelectric OptoGait system (r = 0.99), demonstrating the quality of our reference data. As a use case, the performance of an algorithm originally designed for overground walking was tested on treadmill data using the data pipeline. The accuracy of stride length and stride time estimations was comparable to that reported in other studies with overground data, indicating that the algorithm is equally applicable to treadmill data. The Python source code of the data pipeline is publicly available, and the dataset will be provided by the authors upon request, enabling future evaluations of IMU gait analysis algorithms without the need of recording new data.}, language = {en} } @article{GrischekCaprioglioZhangetal.2022, author = {Grischek, Max and Caprioglio, Pietro and Zhang, Jiahuan and Pena-Camargo, Francisco and Sveinbjornsson, Kari and Zu, Fengshuo and Menzel, Dorothee and Warby, Jonathan and Li, Jinzhao and Koch, Norbert and Unger, Eva and Korte, Lars and Neher, Dieter and Stolterfoht, Martin and Albrecht, Steve}, title = {Efficiency Potential and Voltage Loss of Inorganic CsPbI2Br Perovskite Solar Cells}, series = {Solar RRL}, volume = {6}, journal = {Solar RRL}, number = {11}, publisher = {Wiley-VCH}, address = {Weinheim}, issn = {2367-198X}, doi = {10.1002/solr.202200690}, pages = {12}, year = {2022}, abstract = {Inorganic perovskite solar cells show excellent thermal stability, but the reported power conversion efficiencies are still lower than for organic-inorganic perovskites. This is mainly caused by lower open-circuit voltages (V(OC)s). Herein, the reasons for the low V-OC in inorganic CsPbI2Br perovskite solar cells are investigated. Intensity-dependent photoluminescence measurements for different layer stacks reveal that n-i-p and p-i-n CsPbI2Br solar cells exhibit a strong mismatch between quasi-Fermi level splitting (QFLS) and V-OC. Specifically, the CsPbI2Br p-i-n perovskite solar cell has a QFLS-e center dot V-OC mismatch of 179 meV, compared with 11 meV for a reference cell with an organic-inorganic perovskite of similar bandgap. On the other hand, this study shows that the CsPbI2Br films with a bandgap of 1.9 eV have a very low defect density, resulting in an efficiency potential of 20.3\% with a MeO-2PACz hole-transporting layer and 20.8\% on compact TiO2. Using ultraviolet photoelectron spectroscopy measurements, energy level misalignment is identified as a possible reason for the QFLS-e center dot V-OC mismatch and strategies for overcoming this V-OC limitation are discussed. This work highlights the need to control the interfacial energetics in inorganic perovskite solar cells, but also gives promise for high efficiencies once this issue is resolved.}, language = {en} } @inproceedings{KlippertStolpmannGrumetal.2023, author = {Klippert, Monika and Stolpmann, Robert and Grum, Marcus and Thim, Christof and Gronau, Norbert and Albers, Albert}, title = {Knowledge transfer quality improvement}, series = {Procedia CIRP}, volume = {119}, booktitle = {Procedia CIRP}, publisher = {Elsevier}, address = {Amsterdam}, issn = {2212-8271}, doi = {10.1016/j.procir.2023.02.171}, pages = {919 -- 925}, year = {2023}, abstract = {Developing a new product generation requires the transfer of knowledge among various knowledge carriers. Several factors influence knowledge transfer, e.g., the complexity of engineering tasks or the competence of employees, which can decrease the efficiency and effectiveness of knowledge transfers in product engineering. Hence, improving those knowledge transfers obtains great potential, especially against the backdrop of experienced employees leaving the company due to retirement, so far, research results show, that the knowledge transfer velocity can be raised by following the Knowledge Transfer Velocity Model and implementing so-called interventions in a product engineering context. In most cases, the implemented interventions have a positive effect on knowledge transfer speed improvement. In addition to that, initial theoretical findings describe factors influencing the quality of knowledge transfers and outline a setting to empirically investigate how the quality can be improved by introducing a general description of knowledge transfer reference situations and principles to measure the quality of knowledge artifacts. To assess the quality of knowledge transfers in a product engineering context, the Knowledge Transfer Quality Model (KTQM) is created, which serves as a basis to develop and implement quality-dependent interventions for different knowledge transfer situations. As a result, this paper introduces the specifications of eight situation-adequate interventions to improve the quality of knowledge transfers in product engineering following an intervention template. Those interventions are intended to be implemented in an industrial setting to measure the quality of knowledge transfers and validate their effect.}, language = {en} } @misc{PanzerBenderGronau2021, author = {Panzer, Marcel and Bender, Benedict and Gronau, Norbert}, title = {Deep reinforcement learning in production planning and control}, series = {Zweitver{\"o}ffentlichungen der Universit{\"a}t Potsdam : Wirtschafts- und Sozialwissenschaftliche Reihe}, journal = {Zweitver{\"o}ffentlichungen der Universit{\"a}t Potsdam : Wirtschafts- und Sozialwissenschaftliche Reihe}, issn = {2701-6277}, doi = {10.25932/publishup-60572}, url = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus4-605722}, pages = {13}, year = {2021}, abstract = {Increasingly fast development cycles and individualized products pose major challenges for today's smart production systems in times of industry 4.0. The systems must be flexible and continuously adapt to changing conditions while still guaranteeing high throughputs and robustness against external disruptions. Deep reinforcement learning (RL) algorithms, which already reached impressive success with Google DeepMind's AlphaGo, are increasingly transferred to production systems to meet related requirements. Unlike supervised and unsupervised machine learning techniques, deep RL algorithms learn based on recently collected sensorand process-data in direct interaction with the environment and are able to perform decisions in real-time. As such, deep RL algorithms seem promising given their potential to provide decision support in complex environments, as production systems, and simultaneously adapt to changing circumstances. While different use-cases for deep RL emerged, a structured overview and integration of findings on their application are missing. To address this gap, this contribution provides a systematic literature review of existing deep RL applications in the field of production planning and control as well as production logistics. From a performance perspective, it became evident that deep RL can beat heuristics significantly in their overall performance and provides superior solutions to various industrial use-cases. Nevertheless, safety and reliability concerns must be overcome before the widespread use of deep RL is possible which presumes more intensive testing of deep RL in real world applications besides the already ongoing intensive simulations.}, language = {en} } @inproceedings{PanzerBenderGronau2021, author = {Panzer, Marcel and Bender, Benedict and Gronau, Norbert}, title = {Deep reinforcement learning in production planning and control}, series = {Proceedings of the Conference on Production Systems and Logistics}, booktitle = {Proceedings of the Conference on Production Systems and Logistics}, publisher = {Institutionelles Repositorium der Leibniz Universit{\"a}t Hannover}, address = {Hannover}, issn = {2701-6277}, doi = {10.15488/11238}, pages = {535 -- 545}, year = {2021}, abstract = {Increasingly fast development cycles and individualized products pose major challenges for today's smart production systems in times of industry 4.0. The systems must be flexible and continuously adapt to changing conditions while still guaranteeing high throughputs and robustness against external disruptions. Deep rein- forcement learning (RL) algorithms, which already reached impressive success with Google DeepMind's AlphaGo, are increasingly transferred to production systems to meet related requirements. Unlike supervised and unsupervised machine learning techniques, deep RL algorithms learn based on recently collected sensor- and process-data in direct interaction with the environment and are able to perform decisions in real-time. As such, deep RL algorithms seem promising given their potential to provide decision support in complex environments, as production systems, and simultaneously adapt to changing circumstances. While different use-cases for deep RL emerged, a structured overview and integration of findings on their application are missing. To address this gap, this contribution provides a systematic literature review of existing deep RL applications in the field of production planning and control as well as production logistics. From a performance perspective, it became evident that deep RL can beat heuristics significantly in their overall performance and provides superior solutions to various industrial use-cases. Nevertheless, safety and reliability concerns must be overcome before the widespread use of deep RL is possible which presumes more intensive testing of deep RL in real world applications besides the already ongoing intensive simulations.}, language = {en} }