Analytische Chemie
Filtern
Erscheinungsjahr
- 2022 (16) (entfernen)
Dokumenttyp
- Vortrag (6)
- Posterpräsentation (5)
- Zeitschriftenartikel (3)
- Sonstiges (2)
Schlagworte
- Bioinformatics (5)
- Mass spectrometry (5)
- Taxonomic analysis (4)
- Virus protoemics (4)
- Graphical models (3)
- ABID (2)
- Antibody identification (2)
- Identity (2)
- Machine learning (2)
- Monoclonal antibody (2)
Organisationseinheit der BAM
- VP.1 eScience (16) (entfernen)
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%.
In diesem Vortrag wird die Perspektive einer digitalen Qualitätsinfrastruktur (QI) auf informatischer Seite vorgestellt. Eine zu entwickelnde QI-Cloud ist die Grundlage einer verteilten IT-Plattform über die digitalisierte Prozesse der QI abgewickelt, Daten sicher vorgehalten und ausgetauscht sowie digitale Zertifikate ausgestellt werden können.
Dazu werden Methoden wie die Distributed Ledger Technologie sowie Smart Standards beschrieben, die das Potential haben, essentielle technologische Bestandteile einer digital transformierten QI zu werden.
We interpret solving the multi-vehicle routing problem as a team Markov game with partially observable costs. For a given set of customers to serve, the playing agents (vehicles) have the common goal to determine the team-optimal agent routes with minimal total cost. Each agent thereby observes only its own cost. Our multi-agent reinforcement learning approach, the so-called multi-agent Neural Rewriter, builds on the single-agent Neural Rewriter to solve the problem by iteratively rewriting solutions. Parallel agent action execution and partial observability require new rewriting rules for the game. We propose the introduction of a so-called pool in the system which serves as a collection point for unvisited nodes. It enables agents to act simultaneously and exchange nodes in a conflict-free manner. We realize limited disclosure of agent-specific costs by only sharing them during learning. During inference, each agents acts decentrally, solely based on its own cost. First empirical results on small problem sizes demonstrate that we reach a performance close to the employed OR-Tools benchmark which operates in the perfect cost information setting.
Probability based taxonomic profiling of viral and microbiome samples using PepGM and Unipept
(2022)
In mass spectrometry based proteomics, protein homology leads to many shared peptides within and between species. This complicates
taxonomic inference in samples of unknown taxonomic origin. PepGM uses a graphical model for taxonomic profiling of viral proteomes and
metaproteomic datasets providing taxonomic confidence scores. To build the graphical model, a list of potentially present taxa needs to be
inferred. To this end, we integrate Unipept, which enables the fast querying of potentially present taxa. Together, they allow for taxonomic
inference with statistically sound confidence scores.
In mass spectrometry based proteomics, protein homology leads to
many shared peptides within and between species. This complicates
taxonomic inference. inference. We introduce PepGM, a graphical model for taxonomic profiling of viral proteomes and metaproteomic datasets.
Using the graphical model, our approach computes statistically sound
scores for taxa based on peptide scores from a previous database
search, eliminating the need for commonly used heuristics. heuristics.
In mass spectrometry based proteomics, protein homology leads to
many shared peptides within and between species. This complicates
taxonomic inference. inference. We introduce PepGM, a graphical model for taxonomic profiling of viral proteomes and metaproteomic datasets.
Using the graphical model, our approach computes statistically sound
scores for taxa based on peptide scores from a previous database
search, eliminating the need for commonly used heuristics. heuristics.
In mass spectrometry based proteomics, protein homology leads to
many shared peptides within and between species. This complicates
taxonomic inference. inference. We introduce PepGM, a graphical model for taxonomic profiling of viral proteomes and metaproteomic datasets.
Using the graphical model, our approach computes statistically sound
scores for taxa based on peptide scores from a previous database
search, eliminating the need for commonly used heuristics.
In mass spectrometry based proteomics, protein homology leads to
many shared peptides within and between species. This complicates
taxonomic inference. inference. We introduce PepGM, a graphical model for taxonomic profiling of viral proteomes and metaproteomic datasets.
Using the graphical model, our approach computes statistically sound
scores for taxa based on peptide scores from a previous database
search, eliminating the need for commonly used heuristics.
Introduction:
With the introduction of accurate deep learning predictors, spectral matching applications might experience a renaissance in tandem mass spectrometry (MS/MS) driven proteomics. Deep learning models, e.g., Prosit, predict complete MS/MS spectra from peptide sequences and give the unprecedented ability to accurately predict mass spectra that may arise from any given proteome. However, the amount of spectral data is enormous when querying large search spaces, e.g., metaproteomes composed of many different species.
Current spectral library search software, such as SpectraST, is not equipped to meet run time and memory constraints imposed by such large MS/MS databases, covering several millions of peptide spectrum predictions.
Methods:
Inspired by the fragment index data structure that had been introduced with MSFragger, we implement an efficient peak matching algorithm for computing spectral similarity between query and library spectra. Mistle (Metaproteomic index and spectral library search engine) uses index partitioning and SIMD (Single instruction, multiple data) intrinsics, which greatly improves speed and memory efficiency for searching large spectral libraries. Mistle is written in C++20 and highly parallelized.
Results:
We demonstrate the efficiency of Mistle on two predicted spectral libraries for the lab-assembled microbial communities 9MM and SIHUMIx. Compared to the spectral library search engine SpectraST, Mistle shows a >10-fold runtime improvement and is also faster than msSLASH, which uses locality-sensitive hashing. Although Mistle is slower than MSFragger, Mistle‘s memory footprint is an order of magnitude smaller. Furthermore, we find evidence that the spectral matching approach to predicted libraries identifies peptides with higher precision. Mistle detects peptides not found by database search via MSFragger and in turn uncovers unnoticed false discoveries among their matches.
Conclusion:
In this study, we show that predicted spectral libraries can enhance peptide identification for metaproteomics. Mistle provides the means to efficiently search large-scale spectral libraries, highlighted for the microbiota 9MM and SIHUMIx.
Glow discharge optical emission spectroscopy (GD-OES) is a technique for the analysis of solids such as metals, semiconductors, and ceramics. A low-pressure glow discharge plasma is applied in this system, which ‘sputters’ and promotes the sample atoms to a higher energy state. When the atoms return to their ground state, they emit light with characteristic wavelengths, which a spectrometer can detect. Thus, GD-OES combines the advantages of ICP-OES with solid sampling techniques, which enables it to determine the bulk elemental composition and depth profiles. However, direct solid sampling methods such as glow-discharge spectroscopy require reference materials for calibration due to the strong matrix effect.
Reference materials are essential when the accuracy and reliability of measurement results need to be guaranteed to generate confidence in the analysis. These materials are frequently used to determine measurement uncertainty, validate methods, suitability testing, and quality assurance. In addition, they guarantee that measurement results can be compared to recognized reference values. Unfortunately, the availability of certified reference materials suited to calibrate all elements in different matrix materials is limited. Therefore various calibration strategies and the preparation of traceable matrix-matched calibration standards will be discussed.
Machine learning is an essential component of the growing field of data science. Through statistical methods, algorithms are trained to make classifications or predictions, uncovering key insights within data mining projects. Therefore, it was tried in our work to combine GD-OES with machine learning strategies to establish a new and robust calibration model, which can be used to identify the elemental composition and concentration of metals from a single spectrum. For this purpose, copper reference materials from different manufacturers, which contain various impurity elements, were investigated using GD-OES. The obtained spectra information are evaluated with different algorithms (e.g., gradient boosting and artificial neural networks), and the results are compared and discussed in detail.
Im Auftrag des Bundesministeriums für Wirtschaft und Klimaschutz haben DIN und DKE im Januar 2022 die Arbeiten an der zweiten Ausgabe der Deutschen Normungsroadmap Künstliche Intelligenz gestartet. In einem breiten Beteiligungsprozess und unter Mitwirkung von mehr als 570 Fachleuten aus Wirtschaft, Wissenschaft, öffentlicher Hand und Zivilgesellschaft wurde damit der strategische Fahrplan für die KI-Normung weiterentwickelt. Koordiniert und begleitet wurden diese Arbeiten von einer hochrangigen Koordinierungsgruppe für KI-Normung und -Konformität.
Mit der Normungsroadmap wird eine Maßnahme der KI-Strategie der Bundesregierung umgesetzt und damit ein wesentlicher Beitrag zur „KI – Made in Germany“ geleistet.
Die Normung ist Teil der KI-Strategie und ein strategisches Instrument zur Stärkung der Innovations- und Wettbewerbsfähigkeit der deutschen und europäischen Wirtschaft. Nicht zuletzt deshalb spielt sie im geplanten europäischen Rechtsrahmen für KI, dem Artificial Intelligence Act, eine besondere Rolle.
The amount of data generated worldwide is constantly increasing. These data come from a wide variety of sources and systems, are processed differently, have a multitude of formats, and are stored in an untraceable and unstructured manner, predominantly in natural language in data silos. This problem can be equally applied to the heterogeneous research data from materials science and engineering. In this domain, ways and solutions are increasingly being generated to smartly link material data together with their contextual information in a uniform and well-structured manner on platforms, thus making them discoverable, retrievable, and reusable for research and industry. Ontologies play a key role in this context. They enable the sustainable representation of expert knowledge and the semantically structured filling of databases with computer-processable data triples.
In this perspective article, we present the project initiative Materials-open-Laboratory (Mat-o-Lab) that aims to provide a collaborative environment for domain experts to digitize their research results and processes and make them fit for data-driven materials research and development. The overarching challenge is to generate connection points to further link data from other domains to harness the promised potential of big materials data and harvest new knowledge.
Driven by recent technological advances and the need for improved viral diagnostic applications, mass spectrometry-based proteomics comes into play for detecting viral pathogens accurately and efficiently. However, the lack of specific algorithms and software tools presents a major bottleneck for analyzing data from host-virus samples. For example, accurate species- and strain-level classification of a priori unidentified organisms remains a very challenging task in the setting of large search databases. Another prominent issue is that many existing solutions suffer from the protein inference issue, aggravated because many homologous proteins are present across multiple species. One of the contributing factors is that existing bioinformatic algorithms have been developed mainly for single-species proteomics applications for model organisms or human samples. In addition, a statistically sound framework was lacking to accurately assign peptide identifications to viral taxa. In this presentation, an overview is given on current bioinformatics developments that aim to overcome the above-mentioned issues using algorithmic and statistical methods. The presented methods and software tools aim to provide tailored solutions for both discovery-driven and targeted proteomics for viral diagnostics and taxonomic sample profiling. Furthermore, an outlook is provided on how the bioinformatic developments might serve as a generic toolbox, which can be transferred to other research questions, such as metaproteomics for profiling microbiomes and identifying bacterial pathogens.
In view of the increasing digitization of research and the use of data-intensive measurement and analysis methods, research institutions and their staff are faced with the challenge of documenting a constantly growing volume of data in a comprehensible manner, archiving them for the long term, and making them available for discovery and re-use by others in accordance with the FAIR principles. At BAM, we aim to facilitate the integration of research data management (RDM) strategies during the whole research cycle from the creation and standardized description of materials datasets to their publication in open repositories. To this end, we present the BAM Data Store, a central system for internal RDM that fulfills the heterogenous demands of materials science and engineering labs. The BAM Data Store is based on openBIS, an open-source software developed by the ETH Zurich that has originally been created for life science laboratories but that has since been deployed in a variety of research domains. The software offers a browser-based user interface for the digital representation of lab inventory entities (e.g., samples, chemicals, instruments, and protocols) and an electronic lab notebook for the standardized documentation of experiments and analyses.
To investigate whether openBIS is a suitable framework for the BAM Data Store, we carried out a pilot phase during which five research groups with employees from 16 different BAM divisions were introduced to the software. The pilot groups were chosen to represent a diverse array of domain use cases and RDM requirements (e.g., small vs big data volume, heterogenous vs structured data types) as well as varying levels of prior IT knowledge on the users’ side.
Overall, the results of the pilot phase are promising: While the creation of custom data structures and metadata schemas can be time-intensive and requires the involvement of domain experts, the system offers specific benefits in the form of a simplified documentation and automation of research processes, as well as constituting a basis for data-driven analysis. In this way, heterogeneous research workflows in various materials science research domains could be implemented, from the synthesis and characterization of nanomaterials to the monitoring of engineering structures. In addition to the technical deployment and the development of domain-specific metadata standards, the pilot phase also highlighted the need for suitable institutional infrastructures, processes, and role models. An institute-wide rollout of the BAM Data Store is currently being planned.
MALDI-TOF-MS-based identification of monoclonal murine anti-SARS-CoV-2 antibodies within one hour
(2022)
During the SARS-CoV-2 pandemic, many virus-binding monoclonal antibodies have been developed for clinical and diagnostic purposes. This underlines the importance of antibodies as universal bioanalytical reagents. However, little attention is given to the reproducibility crisis that scientific studies are still facing to date. In a recent study, not even half of all research antibodies mentioned in publications could be identified at all. This should spark more efforts in the search for practical solutions for the traceability of antibodies. For this purpose, we used thirty-five monoclonal antibodies against SARS-CoV-2 to demonstrate how sequence-independent antibody identification can be achieved by simple means applied onto the protein. First, we examined the intact and light chain masses of the antibodies relative to the reference material NIST-mAb 8671. Already half of the antibodies could be identified based solely on these two parameters. In addition, we developed two complementary peptide mass fingerprinting methods with MALDI-TOF-MS that can be performed in 45 minutes and had a combined sequence coverage of over 80%. One method is based on the partial acidic hydrolysis of the protein by 5 mM of sulfuric acid at 99 °C. Furthermore, we established a fast way for a tryptic digest without an alkylation step. We were able to show that the distinction of clones is possible simply by a brief visual comparison of the mass spectra. In this work, two clones originating from the same immunization gave the same fingerprints. Later, a hybridoma sequencing confirmed the sequence identity of these sister clones. In order to automate the spectral comparison for larger libraries of antibodies, we developed the online software ABID 2.0 (https://gets.shinyapps.io/ABID/). This open-source software determines the number of matching peptides in the fingerprint spectra. We propose that publications and other documents critically relying on monoclonal antibodies with unknown amino acid sequences should include at least one antibody fingerprint. By fingerprinting an antibody in question, its identity can be confirmed by comparison with a library spectrum at any time and context.
MALDI-TOF-MS-Based Identification of Monoclonal Murine Anti-SARS-CoV-2 Antibodies within One Hour
(2022)
During the SARS-CoV-2 pandemic, many virus-binding monoclonal antibodies have been developed for clinical and diagnostic purposes. This underlines the importance of antibodies as universal bioanalytical reagents. However, little attention is given to the reproducibility crisis that scientific studies are still facing to date. In a recent study, not even half of all research antibodies mentioned in publications could be identified at all. This should spark more efforts in the search for practical solutions for the traceability of antibodies. For this purpose, we used 35 monoclonal antibodies against SARS-CoV-2 to demonstrate how sequence-independent antibody identification can be achieved by simple means applied to the protein. First, we examined the intact and light chain masses of the antibodies relative to the reference material NIST-mAb 8671. Already half of the antibodies could be identified based solely on these two parameters. In addition, we developed two complementary peptide mass fingerprinting methods with MALDI-TOF-MS that can be performed in 60 min and had a combined sequence coverage of over 80%. One method is based on the partial acidic hydrolysis of the protein by 5 mM of sulfuric acid at 99 degrees C. Furthermore, we established a fast way for a tryptic digest without an alkylation step. We were able to show that the distinction of clones is possible simply by a brief visual comparison of the mass spectra. In this work, two clones originating from the same immunization gave the same fingerprints. Later, a hybridoma sequencing confirmed the sequence identity of these sister clones. In order to automate the spectral comparison for larger libraries of antibodies, we developed the online software ABID 2.0. This open-source software determines the number of matching peptides in the fingerprint spectra. We propose that publications and other documents critically relying on monoclonal antibodies with unknown amino acid sequences should include at least one antibody fingerprint. By fingerprinting an antibody in question, its identity can be confirmed by comparison with a library spectrum at any time and context.