bioRxiv Science⌕ Search

Biology subjects

Kalogeropoulos, K.

Publications and source records attributed to Kalogeropoulos, K..

12 recordsLinked to original sources
bioRxiv2026

Learning from tandem mass spectra at scale with a self-supervised foundation model for proteomics

Mass spectrometry-based proteomics increasingly relies on machine learning, yet existing models are trained for defined supervised tasks such as peptide identification, de novo sequencing or fragment intensity prediction, limiting transfer across datasets, instruments and acquisition methods. Here we present InstaNovo-FM, a self-supervised foundation model for bottom-up proteomics trained to reconstruct masked regions of tandem mass spectra. We assemble a diverse training corpus spanning 1.63 billion MS/MS spectra and 184.6 million high-confidence annotations. We train an encoder-only transformer on the annotated tier using a physics-aware masked reconstruction objective. We demonstrate that the InstaNovo-FM embeddings encode fundamental experimental and biological properties, including fragmentation method, sequence properties and post-translational modifications, without requiring peptide labels. Furthermore, this foundation model directly enables diverse downstream applications, including de novo peptide sequencing, database-free identification and analytical run classification. InstaNovo-FM establishes a unified representation space for peptide fragmentation spectra, enabling robust transferability across the proteomics ecosystem.

bioinformatics↗
bioRxiv2026

Mapping Tissue-Specific Protease Dynamics in the Pig Respiratory Tract During Influenza A Virus Infection

Influenza A virus (IAV) is a zoonotic pathogen capable of infecting diverse avian and mammalian hosts, causing seasonal epidemics and occasional global pandemics in humans. Viral entry requires proteolytic activation of hemagglutinin (HA). While serine proteases such as TMPRSS2, and HAT are known HA activators, the respiratory tract harbours additional proteases whose contributions to infection remain unclear. Dysregulation of these proteases can enhance viral replication, tissue damage, and inflammation, highlighting the need for a systems-level view of the proteolytic landscape. Here, we use high-throughput, proteome-wide proteomics and N-terminomics to identify 112 host proteases across the nasal mucosa, trachea, and lung. We monitor and validate 28 proteases with targeted proteomics and microfluidic qPCR, representing a comprehensive degradome analysis in the respiratory tract of the highly translational pig model of influenza infection. We show that protease abundance and activity were highly tissue-specific: while the nasal mucosa showed selective activation of broad- and narrow-specificity proteases alongside robust antiviral responses, the trachea exhibited modest modulation with subtle shifts in protease-inhibitor balance, and the lung maintained predominantly active proteases despite lower viral loads but severe tissue damage, indicative of immune-mediated pathology. Sequence motif analysis revealed distinct cleavage preferences across tissues, indicating differential protease processing across the studied respiratory tissues in antiviral pathways, antigen processing, and tissue remodelling. Several identified proteases, including ST14, KLKB1, PRSS8, and LGMN, were increased and functionally active upon infection, suggesting roles in viral processing and host immune regulation. Collectively, our results define a spatially organised proteolytic network that shapes tissue-specific antiviral host responses and contributes to H1N1 influenza pathogenesis.

immunology↗
bioRxiv2025

Generalizable direct protein sequencing with InstaNexus

Protein-based therapeutics, such as antibodies and nanobodies, are not encoded in reference genomes, challenging their accurate characterization via standard proteomics. Current methods rely on indirect inference, fragmented outputs, and labor-intensive workflows, which hinder functional insights and routine application. Here, we present a generalizable, end-to-end workflow for direct protein sequencing, combining streamlined sample preparation, AI-driven de novo peptide sequencing, and tailored assembly to reconstruct contiguous protein sequences. A novel composite scoring framework prioritises longer assemblies and coverage, enhancing accuracy and reducing ambiguity. Validation across diverse protein modalities demonstrates its utility and ability to robustly sequence functionally critical regions of selected proteins. This workflow represents an advance in precision proteomics with promising applications in therapeutic discovery, immune profiling, and protein science.

bioinformatics↗
bioRxiv2025

Pathological meprin α expression associated with degradation of dermokine drives a psoriasis-like skin phenotype in a genetic mouse model

Keratinocyte proliferation and differentiation is regulated via proteolytic networks. Dysregulation of proteases within these networks can cause hyperproliferative and inflammatory skin disorders. In healthy skin the metalloprotease meprin is localized in the stratum basale. In contrast, in wound healing tissue and psoriatic lesions increased meprin levels are found in the upper epidermal layers. We developed a transgenic mouse model for inducible expression of pathological meprin levels (K5M) to investigate its epidermal degradome and identify molecular links to keratinocyte proliferation and skin inflammation. K5M mice developed a severe skin phenotype characterized by hyperkeratosis, acanthosis and parakeratosis accompanied by increased transepidermal water loss and a strong inflammatory response within six days after induction of meprin overexpression. Histological and molecular analyses showed that increasing meprin expression correlates with meprin activity and keratinocyte hyperproliferation. Proteomics analyses revealed massive changes in proteins associated with keratinocyte differentiation and epidermal barrier integrity already three days after induction. N-terminomics data indicated a dominant chymotryptic activity with elevated proteolytic turnover of proteins associated with the cytoskeleton, cellular stress responses and cell adhesion. By filtering for cleavage sites that match with the specificity of meprin , we identified highly elevated dermokine-derived peptides. Subsequent mass spectrometric analyses validated dermokine as a novel substrate of meprin and identified the cleavage site, which is highly conserved in mammals. Based on the striking similarities with the phenotype reported for dermokine {beta}{gamma}-/- mice, we propose meprin as a central regulator of keratinocyte proliferation and leukocyte recruitment by proteolytic inactivation of dermokine. Hence, pathological meprin activity could be a driver of hyperproliferative, inflammatory skin disorders like psoriasis vulgaris.

molecular biology↗
bioRxiv2025

InstaNovo-P: A de novo peptide sequencing model for phosphoproteomics

Phosphorylation, a crucial post-translational modification (PTM), plays a central role in cellular signaling and disease mechanisms. Mass spectrometry-based phosphoproteomics is widely used for system-wide characterization of phosphorylation events. However, traditional methods struggle with accurate phosphorylated site localization, complex search spaces, and detecting sequences outside the reference database. Advances in de novo peptide sequencing offer opportunities to address these limitations, but have yet to become integrated and adapted for phosphoproteomics datasets. Here, we present InstaNovo-P, a phosphorylation specific version of our transformer-based InstaNovo model, fine-tuned on extensive phosphoproteomics datasets. InstaNovo-P significantly surpasses existing methods in phosphorylated peptide detection and phosphorylated site localization accuracy across multiple datasets, including complex experimental scenarios. Our model robustly identifies peptides with single and multiple phosphorylated sites, effectively localizing phosphorylation events on serine, threonine, and tyrosine residues. We experimentally validate our model predictions by studying FGFR2 signaling, further demonstrating that InstaNovo-P uncovers phosphorylated sites previously missed by traditional database searches. These predictions align with critical biological processes, confirming the models capacity to yield valuable biological insights. InstaNovo-P adds value to phosphoproteomics experiments by effectively identifying biologically relevant phosphorylation events without prior information, providing a powerful analytical tool for the dissection of signaling pathways.

bioinformatics↗
bioRxiv2024

Systematic benchmarking of mass spectrometry-based antibody sequencing reveals methodological biases

The circulating antibody repertoire is crucial for immune protection, holding significant immunological and biotechnological value. While bottom-up mass spectrometry (MS) is the most widely used proteomics technique for profiling the sequence diversity of circulating antibodies (Ab-seq), it has not been thoroughly benchmarked. We quantified the replicability and robustness of Ab-seq using six monoclonal antibodies with known protein sequences in 70 different combinations of concentration and oligoclonality, both with and without polyclonal serum IgG background. Each combination underwent four protease treatments and was analyzed across four experimental and three technical replicates, totaling 3,360 LC-MS/MS runs. We quantified the dependence of MS-based Ab-seq identification on antibody sequence, concentration, protease, background signal diversity, and bioinformatics setups. Integrating the data from experimental replicates, proteases, and bioinformatics tools enhanced antibody identification. De novo peptide sequencing showed similar performance to database-dependent methods for higher antibody concentrations, but de novo antibody reconstruction remains challenging. Our work provides a foundational resource for the field of MS-based antibody profiling.

immunology↗
bioRxiv2023

CLIPPER 2.0: Peptide level annotation and data analysis for positional proteomics

Positional proteomics methodologies have transformed protease research, and have brought mass spectrometry (MS)-based degradomics studies to the forefront of protease characterization and system-wide interrogation of protease signaling. Considerable advancements in sensitivity and throughput of liquid chromatography (LC)-MS/MS instrumentation enable generation of enormous positional proteomics datasets of natural and protein termini and neo-termini of cleaved protease substrates. However, such progress has not been observed to the same extent in data analysis and post-processing steps, which arguably constitute the largest bottleneck in positional proteomics workflows. Here, we present a computational tool, CLIPPER 2.0, that builds on prior algorithms developed for MS-based protein termini analysis, facilitating peptide level annotation and data analysis. CLIPPER 2.0 can be used with several sample preparation workflows and proteomics search algorithms, and enables fast and automated database information retrieval, statistical and network analysis, as well as visualization of terminomic datasets. We demonstrate our tool by analyzing GluC and MMP9 cleavages in HeLa lysates. CLIPPER 2.0 is available at https://github.com/UadKLab/CLIPPER-2.0. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=192 SRC="FIGDIR/small/569335v1_ufig1.gif" ALT="Figure 1"> View larger version (83K): org.highwire.dtl.DTLVardef@16d603borg.highwire.dtl.DTLVardef@1c8d9d0org.highwire.dtl.DTLVardef@1540246org.highwire.dtl.DTLVardef@614c9a_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗
bioRxiv2023

De novo peptide sequencing with InstaNovo: Accurate, database-free peptide identification for large scale proteomics experiments

Bottom-up mass spectrometry-based proteomics is challenged by the task of identifying the peptide that generates a tandem mass spectrum. Traditional methods that rely on known peptide sequence databases are limited and may not be applicable in certain contexts. De novo peptide sequencing, which assigns peptide sequences to the spectra without prior information, is valuable for various biological applications; yet, due to a lack of accuracy, it remains challenging to apply this approach in many situations. Here, we introduce InstaNovo, a transformer neural network with the ability to translate fragment ion peaks into the sequence of amino acids that make up the studied peptide(s). The model was trained on 28 million labelled spectra matched to 742k human peptides from the ProteomeTools project. We demonstrate that InstaNovo outperforms current state-of-the-art methods on benchmark datasets and showcase its utility in several applications. Building upon human intuition, we also introduce InstaNovo+, a multinomial diffusion model that further improves performance by iterative refinement of predicted sequences. Using these models, we could de novo sequence antibody-based therapeutics with unprecedented coverage, discover novel peptides, and detect unreported organisms in different datasets, thereby expanding the scope and detection rate of proteomics searches. Finally, we could experimentally validate tryptic and non-tryptic peptides with targeted proteomics, demonstrating the fidelity of our predictions. Our models unlock a plethora of opportunities across different scientific domains, such as direct protein sequencing, immunopeptidomics, and exploration of the dark proteome. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=130 SRC="FIGDIR/small/555055v3_ufig1.gif" ALT="Figure 1"> View larger version (38K): org.highwire.dtl.DTLVardef@20063org.highwire.dtl.DTLVardef@1679c01org.highwire.dtl.DTLVardef@1332940org.highwire.dtl.DTLVardef@1d9339f_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗
bioRxiv2023

Salmonella-driven Intestinal Edema in Mice is Characterized by Tensed Fibronectin Fibers

Intestinal edema is a common manifestation of numerous gastrointestinal diseases and is characterized by the accumulation of fluid in the interstitial space of the intestinal wall. Technical advances in laser capture microdissection and low-biomass proteomics now allow us to specifically characterize the intestinal edema proteome. Our data identifies a high abundance of antimicrobial factors, but also highlights major contributions from the blood clotting system, extracellular matrix (ECM) and protease - protease inhibitor networks. The ECM is a complex fibrillar network of macromolecules that provides structural and mechanical support to the intestinal tissue. One abundant component of the ECM observed in the Salmonella-driven intestinal edema is the mechanosensitive glycoprotein fibronectin. Using mechanosensitive staining of fibronectin reveals a solely tensed molecular conformation of fibronectin fibers present in the edema, in contrast to other cecal tissue areas, despite the high abundance of proteases able to cleave fibronectin. Co-staining for fibrin(ogen) indicates the formation of a provisional matrix, similar to what is observed in response to skin injury. The absence of low tensional fibronectin fibers and the additional finding of a high number of protease inhibitors in the edema proteome could indicate a critical role of stretched fibronectin fibers in maintaining tissue integrity in the severely inflamed cecum. Understanding these processes may provide valuable functional diagnostic markers of intestinal disease progression. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=104 SRC="FIGDIR/small/550515v1_ufig1.gif" ALT="Figure 1"> View larger version (39K): org.highwire.dtl.DTLVardef@1318c52org.highwire.dtl.DTLVardef@1f35d21org.highwire.dtl.DTLVardef@67370borg.highwire.dtl.DTLVardef@4ec7e6_HPS_FORMAT_FIGEXP M_FIG C_FIG

pathology↗
bioRxiv2023

A Comparative Study of Protein Structure Prediction Tools for Challenging Targets: Snake Venom Toxins

Protein structure determination is a critical aspect of biological research, enabling us to understand protein function and potential applications. Recent advances in deep learning and artificial intelligence have led to the development of several protein structure prediction tools, such as AlphaFold2 and ColabFold. However, their performance has primarily been evaluated on well-characterised proteins, and comparisons using proteins with poor reference templates are lacking. In this study, we evaluated three modelling tools on their prediction of over 1000 snake venom toxin structures with no reference templates. Our findings show that AlphaFold2 (AF2) performed the best across all assessed parameters. We also observed that ColabFold (CF) only scored slightly worse than AF2, while being computationally less intensive. All tools struggled with regions of intrinsic disorder, such as loops and propeptide regions, and performed well in predicting the structure of functional domains. Overall, our study highlights the importance of exercising caution when working with proteins that have poor reference templates, are large, and contain flexible regions. Nonetheless, leveraging computational structure prediction tools can provide valuable insights into the modelling of protein interactions with different targets and reveal potential binding sites, active sites, and conformational changes, as well as into the design of potential molecular binders for reagent, diagnostic, or therapeutic purposes. StatementRecent advances in machine learning have led to the development of new protein structure prediction tools. However, these tools have mainly been tested on well-known proteins and their performance on proteins without known templates is unclear. This study evaluated the performance of three tools on over 1000 snake venom toxins. We found that while caution is required when studying poorly characterised proteins, these tools offer valuable opportunities to understand protein function and applications.

bioinformatics↗
bioRxiv2021

Black-necked spitting cobra (Naja nigricollis) phospholipases A2 cause Trypanosoma brucei death by blocking endocytosis through the flagellar pocket

African trypanosomes, such as Trypanosoma brucei, are flagellated protozoa which proliferate in mammals and cause a variety of diseases in people and animals. In a mammalian host, the external face of the African trypanosome plasma membrane is covered by a densely packed coat formed of variant surface glycoprotein (VSG), which counteracts the host adaptive immune response by antigenic variation. The VSG is attached to the external face of the plasma membrane by covalent attachment of the C-terminus to a glycosylphosphatidylinositol. As the trypanosome grows, newly synthesised VSG is added to the plasma membrane by vesicle fusion to the flagellar pocket, the sole location of exo- and endocytosis. Snake venoms contain dozens of components including proteases and phospholipases. Here, we investigated the effect of Naja nigricollis on T. brucei with the aim of describing the response of the trypanosome to hydrolytic attack on the VSG. We found no evidence for VGS hydrolysis however N. nigricollis venom caused: (i) an enlargement of the flagellar pocket, (ii) the Rab11 positive endosomal compartments to adopt an abnormal dispersed localisation, and (iii) a cell cycle arrest prior to cytokinesis. A single protein family, the phospholipases A2s present in N. nigricollis venom, was necessary and sufficient for the effects. This study provides new molecular insight into T. brucei biology and possibly describes mechanisms that could be exploited for T. brucei targeting.

microbiology↗