Biology subjects

Wilhelm, M.

Publications and source records attributed to Wilhelm, M..

At least 19 recordsLinked to original sources

bioRxiv2026

PepPr: De Novo peptide sequencing with peptide priors

Mass spectrometry-based peptide sequencing underpins proteomics. Database search confines to predefined candidates, whereas de novo peptide sequencing (DNPS) struggles with incomplete spectra. Existing hybrid methods use databases only for post hoc correction or retrieve previously identified spectra to guide decoding. We instead combine DNPS with a peptide prior: a language model fitted to a reference proteome, tolerating variants yet uncertain on strongly divergent sequences. Relative to prior-free DNPS, our implementation, PepPr, improved reference peptide identification and improved rather than suppressed variant identification. Using a related proteome as prior, it matched or outperformed homology-driven database search, and its gain over prior-free DNPS grew with relatedness. In antibody sequencing, it improved recovery of peptides spanning the variable-constant boundary, deamidation calls, and assembly, while preserving recall of hypervariable regions inaccessible to database search. Peptide priors thus bring reference knowledge into DNPS without limiting discovery.

bioinformatics↗

bioRxiv2026

Long-Term Dynamics of Esca Disease in Western Switzerland

Agroscope conducted a long-term survey to assess the incidence of esca disease in Western Switzerland between 2001 and 2022 and to characterize its temporal dynamics. This study has four main objectives: (1) to provide a rare and particularly valuable long-term dataset documenting the incidence of esca over more than two decades; (2) to identify potentially important climatic drivers using a recent model-independent variable importance algorithm that quantifies the strength of their association with esca incidence; (3) to develop predictive models of esca incidence using climatic covariates derived from microclimatic weather data; and (4) to investigate the relationship between water balance and regionally aggregated esca incidence. The climatic variables identified by the variable importance algorithm as being most strongly associated with esca incidence at the plot level were primarily related to water balance. The predictive analysis showed that model performance at the plot level was limited. Consistent with previous studies, analysis of the variables contributing to model predictions did not allow us to identify a single factor, or set of factors, that could reliably predict esca incidence. At the regional scale, however, water balance and seasonal vapor pressure deficit (VPD) were strongly associated with esca incidence across years, with drier regions generally showing lower disease incidence. Finally, addressing the initial question that motivated this long-term survey, we observed a slight decreasing trend in esca incidence over the study period. This trend coincided with increasingly dry conditions in Western Switzerland and may therefore be partly associated with changes in regional water balance in the context of climate change. All the datasets and analyses presented here are made available and reproducible.

plant biology↗

bioRxiv2026

Deep brain stimulation of the mesencephalic locomotor centre induces stimulation-dependent behavioural states beyond locomotion

Deep brain stimulation (DBS) of the mesencephalic locomotor region (MLR) has been explored to treat gait disturbances. However, clinical outcomes of MLR-DBS have been disappointing overall, with only modest benefits in some individuals and unwanted effects, including anxiety, reported in others. Preclinical studies in several species, by contrast, show that MLR-DBS can improve locomotor and gait deficits, although defensive behaviours in response to stimulation have also been reported in some species. To investigate this discrepancy, we examined frequency- and amplitude-dependent effects of MLR-DBS in healthy rats, combining semi-supervised and unsupervised behavioural analyses. High-frequency, high-amplitude DBS (80-130 Hz), but not low-frequency stimulation (20-60 Hz), elicited hyperlocomotion intermixed with acute defensive-like behaviours. Low-frequency stimulation instead promoted phases of immobility. Consistently, only high-frequency DBS induced region-specific c-Fos expression in the MLR. Complex behaviours, including hyperlocomotion, rearing, tail rattling, and periods of immobility, were most pronounced in animals with the DBS electrode tip localized to the cuneiform nucleus (CnF) of the MLR. Using AAV tracer constructs for bright labelling of CaMKII-positive neurons and their axons, we identified prominent ascending projections from the CnF to the thalamus, substantia nigra pars compacta, zona incerta, hypothalamus, subthalamic nucleus, and central amygdala (CeA). Retrograde tracing confirmed the CnF-to-CeA projection independently. These findings show that the MLR is anatomically connected to higher-order centres linking motor and defensive networks, with DBS frequency and amplitude critically shaping behavioural outcomes that extend beyond locomotion.

neuroscience↗

bioRxiv2026

Intratumoral expression of JAML on NK cells is controlled by tumor microenvironment and MHC class I interaction

Junctional adhesion molecule-like (JAML) is an adhesion molecule known to promote T cell activation and T cell-mediated tumor rejection. In the current study, we show that JAML expression is enriched on mouse intratumoral NK cells compared with splenic NK cells. JAML+ NK cells were associated with tissue residency and co-expressed the immune checkpoints PD-1 and LAG3. JAML expression could be induced on splenic NK cells by IL-2 and further enhanced by IL-21. JAML levels were inversely correlated with inhibitory signaling, as NK cells expressing self-recognizing Ly49 receptors had reduced JAML expression, suggesting regulation of JAML expression by MHC class I molecules. Interaction with the JAML ligand CXADR also reduced JAML surface expression, indicating that tumor-mediated membrane stripping may represent a mechanism of immunoediting. Although JAML RNA transcripts were detectable in human NK cells, JAML protein was found only intracellularly. Together, these findings identify the JAML-CXADR interaction as a potential regulatory pathway in NK cell-mediated killing of tumors.

immunology↗

bioRxiv2026

GAP-MS: Automated validation of gene predictions using integrated mass ‎spectrometry evidence

1.Accurate genome annotation is fundamental to modern biology, yet distinguishing authentic protein-coding sequences from prediction artifacts remains challenging, particularly in complex plant genomes where automated methods are error-prone and manual curation is rarely feasible due to prohibitive time and costs. Here, we present GAP-MS (Gene model Assessment using Peptides from Mass Spectrometry), an automated proteogenomic pipeline that leverages mass spectrometry evidence to validate the protein-level accuracy of predicted gene models. Applied across 9 major crop species, GAP-MS consistently improved the prediction precision for four widely used gene prediction tools. In addition to filtering likely erroneous models, the pipeline identified hundreds of candidate protein-coding loci absent from current standard reference annotations. These peptide-supported loci were further verified by transcriptional evidence, well-supported functional annotations, and high coding-potential scores. Together, these results demonstrate that direct proteomic evidence can help resolve annotation ambiguities, define high-confidence peptide-supported reference proteomes, and uncover overlooked protein-coding genes, while facilitating the identification of sequences that may require further investigation.

bioinformatics↗

bioRxiv2026

ALK R1275Q mutation drives expansion of SCP-like cells during sympathoadrenal commitment and primes neuroblastoma initiation.

Neuroblastoma (NB) is a pediatric malignancy developing in the sympathoadrenal lineage of the neural crest, characterized by clinical heterogeneity ranging from spontaneous regression to poor outcomes. Activating mutations in the receptor tyrosine kinase anaplastic lymphoma kinase (ALK) are frequently observed in both sporadic and familial NB, yet the functional role of ALK in tumor initiation is not fully understood. Using a patient-derived human induced pluripotent stem cell (iPSC) model of sympathoadrenal development, we show that upon sympathoadrenal lineage commitment, ALK R1275Q, the most common hotspot mutation found in familial NB, sustain a proliferative, immature Schwann cell precursor (SCP)-like cell state with elevated ALK signaling and increased susceptibility to MYCN-driven transformation. While ALK-mutant cells alone did not form tumors in vivo, they cooperated with MYCN to accelerate tumor initiation, suggesting that ALK R1275Q creates a permissive but insufficient state for transformation. These findings define an ALK-driven cell progenitor-like state that facilitates the initiation of NB during embryonal development.

cancer biology↗

bioRxiv2026

Locus coeruleus noradrenaline elicits response profiles distinct from natural arousal in hippocampal neurons and astrocytes

The locus coeruleus (LC)-noradrenaline (NA) system is a central component of the brains response to arousal and stress. Yet how LC activity contributes to the cellular response profiles observed during natural arousal remains unclear. Here, we directly compared natural arousal with selective LC activation in mouse CA1, using physiologically titrated optogenetics, fiber photometry of NA and calcium signals, chronic two-photon imaging, and behavioral monitoring. While natural arousal robustly activated astrocytes, pyramidal cells, and interneurons, direct LC stimulation revealed a striking divergence in cellular response polarity and sensitivity. At levels of LC activation that correspond to arousal and moderate stress, we observed strong and reliable calcium responses in astrocytes, whereas pyramidal neurons and interneurons remained largely unaffected. Only at high-intensity LC stimulation did neurons exhibit a response, characterized by broad population-level inhibition of both pyramidal cells and interneurons, alongside the transient activation of an interneuron subpopulation that occupied distinct laminar positions in CA1. Thus, LC-driven NA release produces cell-specific effects in hippocampal CA1 that are distinct from and, at the population level for neurons even opposite to, cellular dynamics during natural arousal. Together, our results reveal a divergence in how astrocytes and neurons respond to LC-driven NA release and suggest that noradrenergic effects in the hippocampus at moderate levels of arousal and stress are predominantly mediated by astrocytes.

neuroscience↗

bioRxiv2025

Decoding isozyme-specific substrate recognition in protein arginine deiminases by proteome-wide citrullination mapping

Protein arginine deiminases (PADs) convert arginine to citrulline, altering protein structure and function. Of the five human isozymes, PAD1-4 are catalytically active with distinct tissue-specificities, yet isozyme-specific substrate recognition remains poorly defined. We profiled PAD1-4 substrate landscapes via mass spectrometry, identifying [~]30,000 citrullination sites across [~]5,500 proteins. Only 14% of sites were shared among all four, reflecting distinct sequence preferences: PAD1-2 showed broad specificity, whereas PAD3-4 favored arginines flanked by acidic or glycine residues. These preferences persisted over 10 min-16 h, indicating sequence context rather than temporal dynamics drives specificity. Mutation analysis of eleven PAD4 variants revealed Q346, G403, R639, and H640 as key determinants distinguishing substrate recognition from that of PAD2. This work provides the most comprehensive PADs substrate atlas to date, defining isozyme-specific motifs and molecular determinants, and guiding development of selective inhibitors and probes to interrogate citrullination mechanisms in health and disease.

biochemistry↗

bioRxiv2025

Systematic evaluation of peptide property predictors with explainable AI technique SHAP

Deep learning models are often characterized as black boxes because their layers of various mathematical transformations and activation functions are practically uninterpretable and have little meaning to the user. However, explainable AI methods exist to attribute model output predictions to its inputs. Shapley additive explanations (SHAP) is one such method that directly quantifies the inputs contributions and qualitatively addresses the question of why a model makes a particular prediction. SHAP generally is hampered by its computational cost, which scales very poorly with large inputs. Therefore peptide property predictors that take as input amino acid sequences, roughly between the lengths of 10-30 amino acids, represent ideal systems for applying SHAP. In applying SHAP to models that predict retention time, collisional cross section, peptide flyability, and fragment intensity, we obtain the relative influence that each amino acid has on predicting each property, and furthermore can rationalize the values from the perspective of the amino acids chemistry. Simply correlating the average shapley values per amino acid type over an entire validation dataset has yielded high correlation to published amino acid indices that are highly related to the property being predicted. For instance, our average shapley values for retention time had a 0.973 Pearson correlation with experimentally measured amino acid retention indices at pH 2. In applying SHAP on the Prosit fragment intensity prediction model, there is strong agreement with the mobile proton model, specifically demonstrating the effect of basic amino acids, the proline effect in charge-remote fragmentation, and positively charged amino acids in charge-directed fragmentation. We also use SHAP in a targeted experiment to demonstrate the Pathways in competition behavior of the model, and reveal a very discrete decision process based on basic residues and the fragment charge. While SHAP in this work was applied to models of well understood properties/systems, there is great potential to explain less studied areas of peptide chemistry to provide insights into their mechanisms.

bioinformatics↗

bioRxiv2025

Learning the Unseen: Data-Augmented Deep Learning for PTM Discovery with Prosit-PTM

Post-translational modifications (PTMs) are critical regulators of protein function, yet confidently identifying and localizing PTM sites across proteomes remains a challenging task. Integrating peptide property predictions into spectrum interpretation improves identification performance, but training data enabling zero-shot prediction across diverse PTMs are scarce. Here, we present a major expansion of the ProteomeTools dataset, comprising over 977,000 synthetic peptides, covering 22 PTM-residue combinations. Furthermore we developed Prosit-PTM, a model with chemically-informed encoding and amino acid substitution-based augmentation trained with our novel ground-truth dataset, that achieves accurate zero-shot predictions. Applied to modified peptides, Prosit-PTM enhances PTM-site localization in phosphoproteomics, increases identification of multiply modified peptides in histones, and enables data-driven rescoring for unseen modifications such as HLA peptides. Furthermore, the learned embeddings of amino acids and modifications capture physicochemical relationships underlying PTM-driven HLA presentation. Prosit-PTM is integrated into multiple open-source tools enabling PTM-aware rescoring, site localization, spectral library generation, and beyond.

bioinformatics↗

bioRxiv2025

Modanovo: A Unified Model for Post-Translational Modification-Aware de Novo Sequencing Using Experimental Spectra from In Vivo and Synthetic Peptides

Post-translational modifications (PTMs) play a central role in cellular regulation and are implicated in numerous diseases. Database searching remains the standard for identifying modified peptides from tandem mass spectra, but is hindered by the combinatorial expansion of modification types and sites. De novo peptide sequencing offers an attractive alternative, yet existing methods remain limited to unmodified peptides or a narrow set of PTMs. Here, we curated a large dataset of spectra from endogenous and synthetic peptides from ProteomeTools spanning 19 biologically relevant amino acid-PTM combinations, covering phosphorylation, acetylation, and ubiquitination. We used this dataset to develop Modanovo, an extension of the Casanovo transformer architecture for de novo peptide sequencing. Modanovo achieved robust performance across these amino acid-PTM combinations (median area under the precision-coverage curve 0.92), while maintaining performance on unmodified peptides (0.93), nearly identical to Casanovo (0.94). The model outperformed {pi}-PrimeNovo-PTM and showed increased precision and complementarity to the database search tool MSFragger. Robustness was confirmed across independent datasets, particularly at peptide lengths frequently represented in the curated dataset. Applied to a phosphoproteomics dataset from monkeypox virus-infected cells, Modanovo recovered numerous confident peptides not reported by database search, including new viral phosphosites supported by spectral evidence, thereby demonstrating its complementarity to database-driven identification approaches. These results establish Modanovo as a broadly applicable model for comprehensive de novo sequencing of both modified and unmodified peptides.

bioinformatics↗

bioRxiv2025

Integrating Alternative Fragmentation Techniques into Standard LC-MS Workflows Using a Single Deep Learning Model Enhances Proteome Coverage

We built and characterised a mass spectrometer capable of performing CID (both beam type and resonant type), UVPD, EID and ECD in an automated fashion during an LCMS type experiment. We exploited this ability to generate large datasets through multienzyme deep proteomics experiments for characterisation of these activation techniques. As a further step, motivated by the complexity generated by these dissociation techniques, we developed a single Prosit deep learning model for fragment ion intensity prediction covering all of these techniques. The generated model has been made publicly available and has been utilised in FragPipe within its MSBooster module. Rescoring allowed both data-dependent acquisition (DDA) and data-independent acquisition (DIA) to achieve on average more than 10% increase in protein identifications across all dissociation techniques and enzymatic digestions. We demonstrate that these alternative fragmentation approaches can now be used within standard data analysis pipelines and can produce data competitive to CID in terms of eficiency, but in the cases of EID and UVPD with far richer and more comprehensive spectra.

systems biology↗

bioRxiv2025

From Hype to Health Check: Critical Evaluation of Drug Response Prediction Models with DrEval

MotivationLarge-scale drug sensitivity screens have enabled training drug response prediction models based on cancer cell line omics profiles to facilitate personalized medicine. While model performances reported in the literature appear promising, no successful translation to the clinic has been reported. ResultsWe identify six primary obstacles to this: non-reproducible models, data leakage leading to poor generalization, pseudoreplication, biased evaluation metrics, missing ablation studies, and inconsistent viability data reducing comparability across models. Together, these issues lead to overly optimistic performance estimates of state-of-the-art models and make it challenging to track progress in the field. To address this, we present DrEval, a pipeline for unbiased and biologically meaningful evaluation of cancer drug response models. It includes baseline and literature models with consistent hyperparameter tuning, statistically sound evaluations, and cross-study benchmarks. DrEval enables ablation studies and publication-ready visualizations. It allows researchers to focus on model development without implementing their own evaluation protocol. We find that deep learning models barely outperform a naive model predicting the mean drug and cell line effects, while no complex model significantly outperforms properly tuned tree-based ensemble baselines in relevant settings. We advocate making our pipeline a standard benchmark for cancer drug response prediction, ensuring a clinically relevant and robust assessment. Availability and implementationDrEval consists of a Python package, available on PyPI (drevalpy) and GitHub (github.com/daisybio/drevalpy), and an accompanying nf-core pipeline (github.com/nf-core/drugresponseeval). All data is available on Zenodo (DOI: 10.5281/zenodo.12633909), preprocessing scripts on github.com/daisybio/preprocess_drp_data.

bioinformatics↗

bioRxiv2025

Benchmarking Spectral Library and Database Search Approaches for Metaproteomics Using a Ground-Truth Microbiome Dataset

Mass spectrometry-based metaproteomics, the identification and quantification of thousands of proteins expressed by complex microbial communities, has become pivotal for unraveling functional interactions within microbiomes. However, metaproteomics data analysis encounters many challenges, including the search of tandem mass spectra against a protein sequence database using proteomics database search algorithms. We used a ground-truth dataset to assess a spectral library searching method against established database searching approaches. Mass spectrometry data collected by data-dependent acquisition (DDA-MS) was analyzed using database searching approaches (MaxQuant and FragPipe), as well as using Scribe with Prosit predicted spectral libraries. We used FASTA databases that included protein sequences from microbial species present in the ground-truth dataset along with background protein sequences, to estimate error rates and assess the effects on detection, peptide-spectral match quality, and quantification. Using the Scribe search engine resulted in more proteins detected at a 1% false discovery rate (FDR) compared to MaxQuant or FragPipe, while FragPipe detected more peptides verified by PepQuery. Scribe was able to detect more low-abundance proteins in the microbiome dataset and was more accurate in quantifying the microbial community composition. This research provides insights and guidance for metaproteomics researchers aiming to optimize results in their analysis of DDA-MS data.

bioinformatics↗

bioRxiv2025

High-throughput chemical proteomics workflow for profiling protein citrullination dynamics

Citrullination is a post-translational modification implicated in autoimmune and inflammatory diseases, yet its low abundance and lack of effective enrichment tools have limited proteome-wide analysis. Here, we developed a robust chemical proteomics workflow with improved specificity and throughput. This method builds upon glyoxal-based derivatization and incorporates a cleavable biotin linker for efficient peptide enrichment, release, and identification via mass spectrometry. Benchmarking across biological systems demonstrated a >10-fold increase in the detection of citrullinated peptides (> 150-fold increase in intensity) at sub-0.1% abundance. Applying the workflow to mouse brain tissue and human primary neutrophils revealed dynamics and condition-specific changes in the citrullinome, including previously uncharacterized sites and regulatory processes. Notably, extensive citrullination of linker histone H1 and structural proteins such as lamin B1 in ionomycin-activated neutrophils suggests broad remodeling of cell architecture via citrullination. This workflow enables proteome-wide mapping of citrullination sites and facilitates its study across diverse biological contexts.

biochemistry↗

bioRxiv2025

Pairwise Attention: Leveraging Mass Differences to Enhance De Novo Sequencing of Mass Spectra

A fundamental challenge in mass spectrometry-based proteomics is determining which peptide generated a given MS2 spectrum. Peptide sequencing typically relies on matching spectra against a known sequence database, which in some applications is not available. Deep learning-based de novo sequencing can address this limitation by directly predicting peptide sequences from MS2 data. We have seen the application of the transformer architecture to de novo sequencing produce state-of-the-art results on the so-called nine-species benchmark. In this study, we propose an improved transformer encoder inspired by the heuristics used in the manual interpretation of spectra. We modify the attention mechanism with a learned bias based on pairwise mass differences, termed Pairwise Attention (PA). Adding PA improves average peptide precision at 100% coverage by 12.7% (5.9 percentage points) over our base transformer on the original nine-species benchmark. We have also achieved a 7.4% increase over the previously published model Casanovo. Our MS2 encoding strategy is largely orthogonal to other transformer-based models encoding MS2 spectra, enabling straightforward integration into existing deep-learning approaches. Our results show that integrating domain-specific knowledge into transformers boosts de novo sequencing performance.

bioinformatics↗

bioRxiv2024

Prosit-XL: enhanced cross-linked peptide identification by accurate fragment intensity prediction to study protein-protein interactions and protein structures

It has been shown that integrating peptide property predictions such as fragment intensity into the scoring process of peptide spectrum match can greatly increase the number of confidently identified peptides compared to using traditional scoring methods. Here, we introduce Prosit-XL, a robust and accurate fragment intensity predictor covering the cleavable (DSSO/DSBU) and non-cleavable cross-linkers (DSS/BS3), achieving high accuracy on various holdout sets with consistent performance on external datasets without fine-tuning. Due to the complex nature of false positives in XL-MS, a novel approach to data-driven rescoring was developed that benefits from Prosit-XLs predictions while limiting the overestimation of the false discovery rate (FDR). We first evaluated this approach using two ground truth datasets that demonstrate the accurate and precise FDR estimation. Second, we applied Prosit-XL on a proteome-scale dataset, demonstrating an up to [~]3.4-fold improvement in PPI discovery compared to classic approaches. Finally, Prosit-XL was used to increase the coverage and depth of a spatially resolved interactome map of intact human cytomegalovirus virions, leading to the discovery of previously unobserved interactions between human and cytomegalovirus proteins.

bioinformatics↗

bioRxiv2024

Exploring Mechanisms of Action in Combinatorial Therapy through Stability/Solubility Alterations: Advancing AML Treatment

Acute myeloid leukemia (AML) remains a formidable clinical challenge due to genetic heterogeneity, high relapse rates, and toxicities associated with conventional chemotherapies. Rationally designed drug combinations offer improved efficacy, yet their selection is often empirical and lacks molecular mechanistic understanding. Here, we present CoPISA workflow (Proteome Integral Solubility/Stability Alteration Analysis for Combinations), a high-throughput proteomics workflow that captures protein solubility/stability alterations unique to combinatorial drug treatments, revealing mechanisms unattainable through single-drug analyses. Applying CoPISA to two rationally designed AML drug pairs, LY3009120-sapanisertib (LS) and ruxolitinib-ulixertinib (RU), we mapped primary (lysate) and secondary (living cell) protein target landscapes. Notably, our analysis uncovered an emergent mechanistic principle, "conjunctional targeting" (i.e., conjunctional inhibition), wherein cooperative drug actions induce treatment-specific targets not achievable individually, analogous to an AND-gate logic model. LS-specific AND-gate proteins converged on SUMOylation, chromatin condensation, and VEGF-linked adhesion, while RU-specific targets disrupted DNA-damage checkpoints, mitochondrial bioenergetics, and RNA-splicing machinery, collectively implicating synthetic-lethal vulnerabilities. Additionally, the post-translational modifications (PTMs) profiling of differential soluble proteins confirms several combination-induced modifications (e.g., acetylation, dimethylation, phosphorylation) on key AML proteins, such as NPM1. Network interrogation of AML-associated proteins showed that a high percentage of targeted proteins are unique to the combinations, including frequently mutated drivers DNMT3A, NPM1, and TP53. CoPISA exposes how drug pairs enact multi-axis pressure on AML cells through conjunctional targeting, a mechanistic layer beyond classical synergy. By pinpointing combination-exclusive protein targets and signaling pathways, CoPISA provides a blueprint for precision-guided regimen design in AML and other heterogeneous cancers. Data are available via ProteomeXchange with identifier PXD066812.

biochemistry↗