Biology subjects

Muth, T.

Publications and source records attributed to Muth, T..

16 recordsLinked to original sources

bioRxiv2026

ProteoDUDes: Taxonomic profiling for metaproteomics with false positive reduction

Metaproteomics is the investigation of the protein composition of multi-organism samples. While metagenomics answers the question which organisms are present in a sample, metaproteomics additionally answers the question which organisms are active. State-of-the-art tools for annotating proteomic data with taxonomic information (e.g. Unipept, DIAMOND) do not control the false taxonomic identification rate, which can lead to incorrect results and thus incorrect interpretations, as we demonstrate with examples. ProteoDUDes processes the results from popular sequence annotation tools so that the proportion of true identifications in the result is at as high as or higher than in the the compared tools. We evaluate ProteoDUDes on simulated data and experimental mock community data. Our results indicate that ProteoDUDes has the same error rate as other tools on simulated data and half the error rate on the experimental mock community data. This allows more accurate statements to be made about which organisms are functionally active in a complex sample. ProteoDUDes is open-source and available at https://github.com/pirovc/dudes.

bioinformatics↗

bioRxiv2025

Limitations of de novo sequencing in resolving sequence ambiguity

De novo peptide sequencing enables peptide identification from fragmentation spectra without relying on sequence databases. However, incomplete spectra create ambiguity, making unambiguous identification challenging. Recent deep learning advances have produced numerous de novo models that predict sequences and refine peptide-spectrum matches under such conditions. Yet, their relative strengths, weaknesses, and ability to handle spectrum ambiguity remain unclear. Here, we benchmark eight state-of-the-art models on three publicly available proteomics datasets, comparing performance using established metrics and quantifying inter-model agreement. We assess post-processing approaches, including iterative refinement, rescoring, and reranking, for their ability to improve identification accuracy, and perform an error analysis to identify common mispredictions and their causes. Model performance varied, with considerable overlap of correct identifications. Post-processing yielded no or only modest improvements. Most sequencing errors were model-independent and driven by limited fragment ion coverage, a limitation also observed in database searches with large search spaces.

bioinformatics↗

bioRxiv2025

MegaPX: fast and space-efficient peptide assignment method using IBF-based multi-indexing

MotivationA central problem for metaproteomic analysis is the often-unknown taxonomic composition of the analyzed microbiomes. Using a database search, the standard approach requires prior knowledge of which proteins and taxa to include in the protein reference database or to use tailored metagenome-derived databases, which are expensive and error-prone in their generation. A possible strategy to circumvent this database search issue is de novo sequencing, where peptide sequences are directly identified from mass spectra. However, these sequences must still be mapped back to potentially extensive databases. Here, alignment-based approaches enable robust and precise results, with the potential drawback of high memory usage and long run times. ResultsWe present MegaPX, a software for rapidly classifying de novo peptide sequences against large protein databases. MegaPX implemented as a C++-based tool, uses an alignment-free, k-mer-based approach as a taxonomic classification method with the possibility of generating mutated reference databases for error-tolerant searching. It uses various algorithms, including interleaved Bloom filters, to efficiently compute approximate membership queries, ensuring fast processing times while querying and indexing large databases in a multi-indexing fashion. We demonstrate the potential of MegaPX by analyzing different samples, including metaproteomics, against extensive reference databases, highlighting its use as a fast screening tool. Availability and implementationMegaPXs source code and all related documentation files are freely available in a GitHub repository (https://github.com/rki-mf2/MegaPX) under the MIT License. ContactMuthT@rki.de Supplementary informationAll supplementary tables and figures are included at the end of the preprint.

bioinformatics↗

bioRxiv2025

Boosting metaproteomics identification rates and taxonomic specificity with MS2Rescore

BackgroundMetaproteomics, the study of the collective proteome within microbial ecosystems, has gained increasing interest over the past decade. However, peptide identification rates in metaproteomics remain low compared to single-species proteomics. A key challenge is the identification sensitivity of current identification algorithms, which were primarily designed for single-species analyses. Addressing this, we evaluated the machine learning-driven MS{superscript 2}Rescore post-processing tool on multiple metaproteomics datasets from diverse microbial environments and benchmark studies. ResultsWe demonstrate that machine learning-driven rescoring outperforms traditional metaproteomics identification workflows. It significantly increases peptide identification rates compared to Sage, which itself already implements basic rescoring. Moreover, it enables lowering the false discovery rate (FDR) to 0.1% with minimal to no sensitivity loss, a substantial improvement over the 1% or 5% FDR thresholds commonly used in metaproteomics, in turn leading to greater confidence in downstream taxonomic annotation. ConclusionsOur findings show that MS{superscript 2}Rescore substantially improves peptide identification sensitivity as well as specificity in metaproteomics, and delivers improved confidence in taxonomic annotation. This advancement results in a more reliable downstream taxonomic analysis, reinforcing the potential of machine learning-based rescoring in metaproteomics research.

bioinformatics↗

bioRxiv2024

The Peptonizer2000: graphical model based taxonomic identifications of metaprotoemic samples

Metaproteomics, the large-scale study of proteins from microbial communities, faces challenges in identifying species due to similarities in protein sequences across different organisms. Current methods often rely on simple counting of matches between proteins and taxa, which can lead to low accuracy. We introduce the Peptonizer2000, a new tool that uses advanced modeling to provide more precise taxonomic identifications along with confidence scores. It combines peptide scores from any proteomic search engine with peptide-to-taxon links from the Unipept database. By applying statistical models, the Peptonizer2000 improves taxonomic resolution and delivers more reliable results. We validate its performance using publicly available datasets, demonstrating its ability to produce high-confidence identifications. Our results suggest that the Peptonizer2000 improves the specificity and confidence of taxonomic assignments in metaproteomics, providing a valuable resource for the study of complex microbial communities.

bioinformatics↗

bioRxiv2024

MultiStageSearch: a multi-step proteogenomic workflow for taxonomic identification of viral proteome samples adressing database bias

The recent years, with the global SARS-Cov-2 pandemic, have shown the importance of strain level identification of viral pathogens. While the gold-standard approach for unkown viral sample identification remains genomics, studies have shown the necessity and advantages of orthogonal experimental approaches such as proteomics, based on proteomic database search methods. The databases required as references for both proteins and genome sequences are known to be biased towards certain taxa, such as pathogenic strains or species, or common model organisms. Aditionally, the proteomic databases are not as comprehensive as the genomic databases. We present MultiStageSearch, an iterative database search approach for the taxonomic identification of viral samples combining proteomic and genomic databases. The potentially present species and strains are inferred using a generalist proteomic reference database. MultiStageSearch then automatically creates a proteogenomic database. This database is further pre-processed byfiltering for duplicates as well as clustering of identical ORFs to address potential bias present in the genomic database. Furthermore, the workflow is independent of the strain level NCBI taxonomy, enabling the inference of strains that are not present in the NCBI taxonomy. We performed a benchmark on several viral samples to demonstrate the performance of the strain level taxonomic inference. The benchmark shows superior performance compared to state of the art methods for untargeted strain level inference using proteomic data while being independent of the NCBI taxonomy at strain level.

bioinformatics↗

bioRxiv2024

Fiora: Local neighborhood-based prediction of compound mass spectra from single fragmentation events

Non-targeted metabolomics holds great promise for advancing precision medicine and facilitating the discovery of novel biomarkers. However, the identification of compounds from tandem mass spectra remains a non-trivial task due to the incomplete nature of spectral reference libraries. Augmenting these libraries with simulated mass spectra can provide the necessary reference to resolve unmatched mass spectra, but remains a difficult undertaking to this day. In this study, we introduce Fiora, an innovative open-source algorithm using graph neural networks to simulate tandem mass spectra in silico. Our objective is to improve fragment intensity prediction with an intricate graph model architecture that facilitates edge prediction, thereby modeling fragment ions as the result of singular bond breaks and their local molecular neighborhood. We evaluate the performance on test data from NIST (2017) and the curated MS-Dial spectral library, as well as compounds from the 2016 and 2022 CASMI challenges. Fiora not only surpasses state-of-the-art fragmentation algorithms, ICEBERG and CFM-ID, in terms of prediction quality, but also predicts additional features, such as retention time and collision cross section. In addition, Fiora demonstrates significant speed improvements through the use of GPUs. This enables rapid (re)scoring of putative compound identifications in non-targeted experiments and facilitates large-scale expansion of spectral reference libraries with accurate spectral predictions. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=105 SRC="FIGDIR/small/590551v1_ufig1.gif" ALT="Figure 1"> View larger version (18K): org.highwire.dtl.DTLVardef@90bf99org.highwire.dtl.DTLVardef@112da33org.highwire.dtl.DTLVardef@b376eeorg.highwire.dtl.DTLVardef@450ee6_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗

bioRxiv2023

gNOMO2: a comprehensive and modular pipeline for integrated multi-omics analyses of microbiomes

BackgroundOver the past few years, the rise of omics technologies has offered an exceptional chance to gain a deeper insight into the structural and functional characteristics of microbial communities. As a result, there is a growing demand for user friendly, reproducible, and versatile bioinformatic tools that can effectively harness multi-omics data to offer a holistic understanding of microbiomes. Previously, we introduced gNOMO, a bioinformatic pipeline specifically tailored to analyze microbiome multi-omics data in an integrative manner. In response to the evolving demands within the microbiome field and the growing necessity for integrated multi-omics data analysis, we have implemented substantial enhancements to the gNOMO pipeline. ResultsHere, we present gNOMO2, a comprehensive and modular pipeline that can seamlessly manage various omics combinations, ranging from two to four distinct omics data types including 16S rRNA gene amplicon sequencing, metagenomics, metatranscriptomics, and metaproteomics. Furthermore, gNOMO2 features a specialized module for processing 16S rRNA gene amplicon sequencing data to create a protein database suitable for metaproteomics investigations. Moreover, it incorporates new differential abundance, integration and visualization approaches, all aimed at providing a more comprehensive toolkit and insightful analysis of microbiomes. The functionality of these new features is showcased through the use of four microbiome multi-omics datasets encompassing various ecosystems and omics combinations. gNOMO2 not only replicated most of the primary findings from these studies but also offered further valuable perspectives. ConclusionsgNOMO2 enables the thorough integration of taxonomic and functional analyses in microbiome multi-omics data, opening up avenues for novel insights in the field of both host associated and free-living microbiome research. gNOMO2 is available freely at https://github.com/muzafferarikan/gNOMO2.

bioinformatics↗

bioRxiv2022

PepGM: A probabilistic graphical model for taxonomic inference of viral proteome samples with associated confidence scores

MotivationInferring taxonomy in mass spectrometry-based shotgun proteomics is a complex task. In multi-species or viral samples of unknown taxonomic origin, the presence of proteins and corresponding taxa must be inferred from a list of identified peptides which is often complicated by protein homology: many proteins do not only share peptides within a taxon but also between taxa. However, correct taxonomic identification is crucial when identifying different viral strains with high sequence homology - considering, e.g., the different epidemiological characteristics of the various strains of SARS-CoV-2. Additionally, many viruses mutate frequently, further complicating the correct assignment of virus proteomic samples. ResultsWe present PepGM, a probabilistic graphical for the taxonomic assignment of virus proteomic samples with strain-level resolution and associated confidence scores. PepGM combines the results of a standard proteomic database search algorithm with belief propagation to calculate the marginal distributions, and thus confidence score, for potential taxonomic assignments. We demonstrate the performance of PepGM using several publicly available virus proteomic datasets, showing its strain-level resolution performance. In two out of eight cases, the taxonomic assignments were only correct on species level, which PepGM clearly indicates by lower confidence scores. Availability and ImplementationPepGM is written in Python and embedded into a Snakemake workflow. Its is available at https://github.com/BAMeScience/PepGM

bioinformatics↗

bioRxiv2022

Mistle: bringing spectral library predictions to metaproteomics with an efficient search index

MotivationDeep learning has moved to the forefront of tandem mass spectrometry-driven proteomics and authentic prediction for peptide fragmentation is more feasible than ever. Still, at this point spectral prediction is mainly used to validate database search results or used for confined search spaces. Fully predicted spectral libraries have not yet been efficiently adapted to large search space problems that often occur in metaproteomics or proteogenomics. ResultsIn this study, we showcase a workflow that uses Prosit for spectral library predictions on two common metaproteomes and implement an indexing and search algorithm, Mistle, to efficiently identify experimental mass spectra within the library. Hence, the workflow emulates a classic protein sequence database search with protein digestion but builds a searchable index from spectral predictions as an in-between step. We compare Mistle to popular search engines, both on a spectral and database search level, and provide evidence that this approach is more accurate than a database search using MSFragger. Mistle outperforms other spectral library search engines in terms of run time and proves to be extremely memory efficient with an 8 to 22-fold decrease in RAM usage. This makes Mistle universally applicable to large search spaces, e.g. covering comprehensive sequence databases of diverse microbiomes. AvailabilityMistle is freely available on GitHub at https://github.com/BAMeScience/Mistle. Contactthilo.muth@bam.de

bioinformatics↗

bioRxiv2022

Current state, existing challenges, and promising progress for de novo sequencing and assembly of monoclonal antibodies

Monoclonal antibodies (mAbs) are biotechnologically produced proteins with various applications in research, therapeutics, and diagnostics. Their ability to recognize and bind to specific molecule structures makes them essential research tools and therapeutic agents. Sequence information of antibodies is helpful for understanding antibody-antigen interactions and ensuring their affinity and specificity. De novo protein sequencing based on mass spectrometry is a useful method to obtain the amino acid sequence of peptides and proteins without a priori knowledge. Deep learning-based approaches have been developed and applied more frequently to increase the accuracy of de novo sequencing. In this study, we evaluated five recently developed de novo sequencing algorithms (Novor, pNovo 3, DeepNovo, SMSNet, and PointNovo) in their ability to identify and assemble antibody sequences. The deep learning-based tools PointNovo and SMSNet showed an increased peptide recall across different enzymes and datasets compared to spectrum-graph-based approaches. We evaluated different error types of de novo peptide sequencing tools and their performance for different numbers of missing cleavage sites, noisy spectra, and peptides of various lengths. We achieved a sequence coverage of 93.15% to 99.07% on the light chains of three different antibody datasets using the de Bruijn assembler ALPS and the predictions from PointNovo. However, low sequence coverage and accuracy on the heavy chains demonstrate that complete de novo protein sequencing remains a challenging issue in proteomics that requires improved de novo error correction, alternative digestion strategies, and hybrid approaches such as homology search to achieve high accuracy on long protein sequences.

bioinformatics↗

bioRxiv2021

yHydra: Deep Learning enables an Ultra Fast Open Search by Jointly Embedding MS/MS Spectra and Peptides of Mass Spectrometry-based Proteomics

Mass spectrometry-based proteomics allows to study all proteins of a sample on a molecular level. However, mass spectra are noisy and contain complex patterns, making them inherently challenging to analyze with purely algorithmic approaches. In terms of the protein sequence landscape, most recent bottom-up MS-based proteomics studies consider either a diverse pool of post-translational modifications, employ large databases - as in metaproteomics or proteogenomics, study multiple isoforms of proteins, include unspecific cleavage sites or even combinations thereof. All this makes peptide and protein identifications challenging due to sheer size of the search space. To cope with this two-sided challenge, i.e. the complexity of real spectra and the search space size, we present a foundation model, called yHydra, that jointly embeds spectra and peptides. This allows us to implement various downstream tasks and search modes in Euclidean space. In particular, we implement an open search which allows to query multiple ten-thousands of spectra against millions of peptides. Furthermore, we implement an error-tolerant search for identifying additional proteoforms that are not included in off-the-shelf reference proteomes. Our foundation model provides meaningful embeddings, as we interpret learned peptide embeddings in comparison to the peptides physico-chemical properties. yHydras open search, assigns delta masses to each identification which allows to unrestrictedly characterize post-translational modifications. The error-tolerant mode of yHydra can be used as post-processing to existing search engines or as a standalone. yHydra is evaluated on several real life data sets for the identification of modified protein sequences and shows up to 25% increase in protein identification at constant false discovery rate compared to the current state-of-the-art. Availability(under MIT license) https://gitlab.com/dacs-hpi/yHydra Contactbernhard.renard@hpi.de

bioinformatics↗

bioRxiv2021

Critical Assessment of Metaproteome Investigation (CAMPI): a Multi-Lab Comparison of Established Workflows

Metaproteomics has matured into a powerful tool to assess functional interactions in microbial communities. While many metaproteomic workflows are available, the impact of method choice on results remains unclear. Here, we carried out the first community-driven, multi-laboratory comparison in metaproteomics: the critical assessment of metaproteome investigation study (CAMPI). Based on well-established workflows, we evaluated the effect of sample preparation, mass spectrometry, and bioinformatic analysis using two samples: a simplified, laboratory-assembled human intestinal model and a human fecal sample. We observed that variability at the peptide level was predominantly due to sample processing workflows, with a smaller contribution of bioinformatic pipelines. These peptide-level differences largely disappeared at the protein group level. While differences were observed for predicted community composition, similar functional profiles were obtained across workflows. CAMPI demonstrates the robustness of present-day metaproteomics research, serves as a template for multi-laboratory studies in metaproteomics, and provides publicly available data sets for benchmarking future developments.

microbiology↗

bioRxiv2020

AHLF: ad hoc learning of fragmentation patterns in mass spectra detects post-translationally modified peptides prior to database search

Mass spectrometry-based proteomics provides a holistic snapshot of the entire protein set of a living cell on a molecular level. Currently, only a few deep learning approaches that involve peptide fragmentation spectra, which represent partial sequence information of proteins, exist. Commonly, these approaches lack the ability to characterize less studied or even unknown patterns in spectra because of their use of explicit domain knowledge. To elevate unrestricted learning from spectra, we introduce AHLF, a deep learning model that is end-to-end trained on 19.2 million spectra from multiple phosphoproteomic data sets. AHLF is interpretable and we show that peak-level feature importances 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 to the current-state-of-the-art on this task. 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%. We expect our approach to directly apply to cell signaling and structural biology which use phosphoproteomic and cross-linking data, but in principal any mass spectrometry based study can benefit from an interpretable, end-to-end trained model like AHLF. Availabilityhttps://gitlab.com/dacs-hpi/ahlf Contactbernhard.renard@hpi.de

bioinformatics↗

bioRxiv2020

Survey of metaproteomics software tools for functional microbiome analysis.

To gain a thorough appreciation of microbiome dynamics, researchers characterize the functional role of expressed microbial genes/proteins. This can be accomplished through metaproteomics, which characterizes the protein complement of the microbiome. Several software tools exist for analyzing microbiomes at the functional level by measuring their combined proteome-level response to environmental perturbations. In this survey, we explore the performance of six available tools, so that researchers can make informed decisions regarding software choice based on their research goals. Tandem mass spectrometry-based proteomic data obtained from dental carie plaque samples grown with and without sucrose in paired biofilm reactors were used as representative data for this evaluation. Microbial peptides from one sample pair were identified by the X! Tandem search algorithm via SearchGUI and subjected to functional analysis using software tools including eggNOG-mapper, MEGAN6, MetaGOmics, MetaProteomeAnalyzer (MPA), ProPHAnE, and Unipept to generate functional annotation through Gene Ontology (GO) terms. Among these software tools, notable differences in functional annotation were detected after comparing differentially expressed protein functional groups. Based on the generated GO terms of these tools we performed a peptide-level comparison to evaluate the quality of their functional annotations. A BLAST analysis against the Universal Protein Knowledgebase revealed that the sensitivity and specificity of functional annotation differed between tools. For example, eggNOG-mapper mapped to the most number of GO terms, while Unipept generated the most precise GO terms. Based on our evaluation, metaproteomics researchers can choose the software according to their analytical needs and developers can use the resulting feedback to further optimize their algorithms. To make more of these tools accessible via scalable metaproteomics workflows, eggNOG-mapper and Unipept 4.0 were incorporated into the Galaxy platform.

bioinformatics↗

bioRxiv2019

TaxIt: An iterative and automated computational pipeline for untargeted strain-level identification using MS/MS spectra from pathogenic samples

Untargeted accurate strain-level classification of a priori unidentified organisms using tandem mass spectrometry is a challenging task. Reference databases often lack taxonomic depth, limiting peptide assignments to the species level. However, the extension with detailed strain information increases runtime and decreases statistical power. In addition, larger databases contain a higher number of similar proteomes.\n\nWe present TaxIt, an iterative workflow to address the increasing search space required for MS/MS-based strain-level classification of samples with unknown taxonomic origin. TaxIt first applies reference sequence data for initial identification of species candidates, followed by automated acquisition of relevant strain sequences for low level classification. Furthermore, proteome similarities resulting in ambiguous taxonomic assignments are addressed with an abundance weighting strategy to improve candidate confidence.\n\nWe apply our iterative workflow on several samples of bacterial and viral origin. In comparison to non-iterative approaches using unique peptides or advanced abundance correction, TaxIt identifies microbial strains correctly in all examples presented (with one tie), thereby demonstrating the potential for untargeted and deeper taxonomic classification. TaxIt makes extensive use of public, unrestricted and continuously growing sequence resources such as the NCBI databases and is available under open-source license at https://gitlab.com/rki_bioinformatics.

bioinformatics↗