Lars Malmström

The paper maps the structural epitope landscape of pneumolysin and identifies a conserved cross-species neutralising site among cholesterol-dependent cytolysins.

This study identifies SpnA's role in inhibiting the complement system by interacting with C5b67, preventing membrane attack complex assembly.

This paper designs an epitope-centric nanoparticle vaccine against Streptococcus pyogenes using structural mass spectrometry and deep learning to enhance immunogenicity.

The paper develops a proteomics strategy to infer the origin of proteins in human plasma using a proteome atlas and RNA/protein atlases, demonstrating its application in disease cohorts.

This paper introduces a novel MS score framework for analyzing polyclonal antibodies, demonstrating its ability to identify antibody subgroups and predict interactions.

This paper develops a combined systems antigenomics-serology workflow to characterize circulating IgG against Group A Streptococcus, revealing antigen repertoire and functional properties of antibodies in healthy and infected individuals.

This study shows that streptolysin O enhances plasminogen to plasmin conversion by interacting with plasminogen and altering its conformation to increase sensitivity to proteolytic processing.

This study uses multi-modal mass spectrometry to identify a conserved protective epitope in Streptolysin O from S. pyogenes, which could inform vaccine and therapeutic development.

This study shows that increasing Fc flexibility in an IgG1-IgG3 hybrid monoclonal antibody enhances opsonic function and protection against group A streptococci.

The paper introduces DPKS, an explainable machine learning framework for identifying proteome states by calculating protein importance in predictive models.

This study uses pharmacoproteomics to analyze treatment responses in a murine sepsis model, identifying distinct response patterns and drug interactions across organs and time.

This paper introduces BINNs, a deep learning method that integrates protein abundance and pathways to improve biomarker discovery and subphenotype classification in proteomics data.

This study develops multienzyme deep learning models for de novo peptide sequencing from mass spectrometry data, showing improved generalizability and sequence coverage across diverse protease digests and species.

This paper introduces GPS, a machine learning-based method for improving peptide and protein identification accuracy and rates in DIA proteomics by using an external dataset, and demonstrates its effectiveness in identifying biomarkers for acute kidney injury.

This study determines the structural interactions between Streptococcus pyogenes M1 protein and human IgG using integrative structural biology, revealing conserved mechanisms to inhibit phagocytosis.

The paper improves a proteogenomics workflow to reveal molecular phenotypes linked to breast cancer appearance, showing differences in protein networks and biomarker abundance.