Biology subjects

Mazurenko, S.

Publications and source records attributed to Mazurenko, S..

15 recordsLinked to original sources

bioRxiv2026

Leakage-Controlled Benchmarking of Variant Effect Predictors and Development of Gradient-Boosting Consensus Model

Single amino acid mutations are among the most common forms of genetic variation, with effects ranging from negligible to severe disruption of protein function. Accurate quantitative evaluation of protein variants therefore represents a key challenge in medicine and protein engineering. Computational variant effect predictors (VEPs) are widely used to prioritize missense variants, yet their fair comparison is hindered by data circularity, benchmarks restricted to human clinical annotations, and limited availability of code and training data, all of which can inflate apparent performance. Here, we reviewed 18 recent and 3 older, widely adopted variant effect predictors for reproducible benchmarking and assembled a rigorously curated, leakage-controlled Benchmark dataset by integrating clinically annotated human variants from ClinVar and Humsavar as well as cross-species variants with functional annotations. The evaluation on this dataset revealed a narrow overall performance range for the benchmarked tools (AUC 0.631-0.758) and a consistent drop from clinical to functional data. Surprisingly, several tools published two decades ago demonstrated comparable performance to mid-ranked modern predictors. Building on these results, we developed a gradient-boosting consensus model, PredictSNPprot, which achieved an AUC of 0.82 on a held-out test set, outperforming the original PredictSNP meta-predictor and the best individual state-of-the-art tool by 13.6 and 5.7 percentage points, respectively. The Benchmark dataset, including the predictions from the evaluated tools, is available at https://doi.org/10.5281/zenodo.23597130 and PredictSNPprot consensus predictor is available at https://github.com/loschmidt/PredictSNPprot.

bioinformatics↗

bioRxiv2026

CascadeMAP: Autonomous Closed-loop Optimization of Enzyme Cascades via Microfluidics, Machine Learning and Agentic AI

Enzyme cascades enable complex biochemical transformations, but their optimization is resource-intensive, requiring navigation through high-dimensional parameter spaces encompassing reaction conditions, enzyme ratios, and buffer composition. Here we introduce CascadeMAP, an autonomous microfluidic platform for closed-loop optimization of enzyme cascades, integrating high-throughput microfluidics with Bayesian optimization and multi-agent AI system. We demonstrate the platform across two cascades: (i) a glycerol detection pathway monitored by fluorescence and (ii) a 1,2,3-trichloropropane degradation pathway monitored by label-free Raman spectroscopy providing orthogonal detection modalities. Bayesian optimization identified optimal conditions three times faster than Design of Experiments. Multi-agent AI system automated hypothesis generation, processing 11 GB of experimental data, pattern recognition, and insight synthesis. Operating without human intervention for 7 days, CascadeMAP processed [~]220,000 reactions across [~]7,400 different conditions. This capability establishes a generalizable framework for the autonomous optimization of enzyme cascades and metabolic pathways and accelerates the development of biocatalytic and synthetic biological systems.

biochemistry↗

bioRxiv2026

ProtXAI: Explainable AI Reveals Structural Determinants of Protein Dynamics

Molecular dynamics simulations provide atomistic views of protein motions, but conventional analyses often struggle with extracting subtle mechanistic insights from complex trajectories. Here, we present an integrated framework, ProtXAI, combining molecular dynamics and explainable artificial intelligence (XAI), to identify residue-level determinants of conformational change across diverse protein systems. By leveraging inter-residue distance dynamics, deep learning, and sequential relevance propagation, the approach captures both local fluctuations and long-range communication pathways within protein structures. We applied this framework to three mechanistically distinct systems: apolipoprotein E4 (ApoE4), staphylokinase (SAK) variants, and an ancestral luciferase. Across these applications, our XAI-based approach recovered experimentally supported dynamic hotspots: ligand-responsive hinges in ApoE4, mutation-dependent flexibility shifts in SAK, and evolutionary redistribution of motions in the luciferase. ProtXAI also revealed additional long-range couplings not accessible to classical analysis. Together, these findings demonstrate that combining molecular dynamics with XAI provides a general and scalable strategy for dissecting protein dynamics and uncovering structural determinants of function, stability, and evolutionary changes without prior bias. This approach thus advances the current methodological repertoire for analysing proteins and their intrinsic properties. HighlightsMolecular dynamics simulations are increasingly accessible, yet scalable tools for comparative analysis remain limited. We demonstrate that machine learning coupled with explainable AI can automatically extract structural determinants of protein dynamics from trajectories. ProtXAI identifies key dynamic regions across diverse scenarios, including comparison of protein variants, understanding ligand modulation, and single-trajectory analysis. ProtXAI enables scalable, unbiased interpretation of long trajectories, providing an alternative to manual, time-intensive analysis.

bioinformatics↗

bioRxiv2026

Investigation of Protein Melting Temperature Prediction with Cross-Method Validation on Biophysical Data

MotivationProtein melting temperature (Tm) prediction accelerates the discovery of thermostable enzymes which are crucial for industrial biotechnology often requiring harsh reaction conditions. Experimental determination of Tm remains labour-intensive and varies across techniques, motivating the development of in silico predictors. Mass-spectrometry datasets such as Meltome Atlas now enable large-scale Tm prediction with models based on deep learning, but model generalisation across diverse experimental datasets has not been systematically tested. ResultsWe evaluated the generalisability of state-of-the-art deep learning approaches and explored ESM-based embeddings for Tm prediction. To this end, we assembled the ProMelt training dataset (45 441 proteins) and five independent biophysics-based validation datasets. Our analysis revealed substantial differences between proteomics- and biophysics-based Tm measurements, highlighting the challenge of cross-domain generalisation. Existing state-of-the-art predictors trained on large-scale proteomics datasets showed reduced performance on biophysics-based validation sets. Our fine-tuned embedding-based models, particularly LoRA-adapted ESM-2 (TmProt 1.0), outperformed state-of-the-art predictors in identifying thermostable proteins (Tm[≥] 60 {degrees}C) across heterogeneous datasets, achieving AUC scores of 0.75-0.77. We also demonstrated that the available models could be used efficiently in the sequence prioritization task. AvailabilityThe TmProt web server is available at https://loschmidt.chemi.muni.cz/tmprot/. Source code and data are available at https://github.com/loschmidt/TmProt.

bioinformatics↗

bioRxiv2026

Uncovering Functional Distant Mutations by Ultra-High-Throughput Screening of Dehalogenases

Conformational dynamics play a central role in enzyme function by controlling substrate access and productive binding. Yet mutations that beneficially modulate these properties are difficult to identify. Here, we used ultrahigh-throughput fluorescence-activated droplet sorting (FADS) with a bulky fluorogenic substrate derived from coumarin (COU-3) to impose steric selection pressure on the haloalkane dehalogenase LinB. Screening a focused library yielded five single substitutions located 11.5-15.5 [A] from the catalytic centre. Variant I138N showed a fourfold increase in catalytic efficiency toward COU-3 through reduced KM and increased kcat, associated with increased cap-domain flexibility and facilitated substrate entry. In contrast, variant P208S markedly reduced substrate inhibition and shifted specificity toward bulkier iodinated haloalkanes by reshaping its tunnel environment. Integrated kinetic and structural analyses revealed that screening with bulky substrates directs selection toward distal regions controlling substrate access and unproductive binding. These findings demonstrate that ultrahigh-throughput FADS can reveal dynamic mechanisms of enzyme adaptation that remain difficult to predict by rational design. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=183 SRC="FIGDIR/small/713925v1_ufig1.gif" ALT="Figure 1"> View larger version (51K): org.highwire.dtl.DTLVardef@782038org.highwire.dtl.DTLVardef@8b43f3org.highwire.dtl.DTLVardef@11a403eorg.highwire.dtl.DTLVardef@6fcaea_HPS_FORMAT_FIGEXP M_FIG C_FIG

biochemistry↗

bioRxiv2026

Data-efficient distal engineering of fluorinase using zero-shot models

Fluorinases have high potential for industrial biofluorination but any applications have been precluded by low catalytic efficiency and resistance to active site engineering. In this work, we employed PRIZM, a computational workflow utilizing an existing low-N dataset and zero-shot models for in silico prediction of activity-enhancing mutations at distal sites. The combination of these predictions with expert opinion led to the identification of 21 fluorinase mutants with enhanced relative activities, while 3 variants showed increased melting temperatures. A mutation in the hexameric interface, K237R, resulted in the largest stability gain, a more than 3.2-fold improvement in catalytic efficiency at 57{degrees}C, and an 8-fold increase in relative activity at 62{degrees}C. These results highlight the potential of distal fluorinase engineering for improving properties required to realize its industrial applications.

bioengineering↗

bioRxiv2026

PRIZM: Combining Low-N Data and Zero-shot Models to Design Enhanced Protein Variants

Machine learning has repeatedly shown the ability to accelerate protein engineering, but many approaches demand large amounts of robust, high-quality training data as well as substantial computational expertise. While large pre-trained models can function as zero-shot proxies for predicting variant effects, selecting the best model for a given protein property is often non-trivial. Here, we introduce Protein Ranking using Informed Zero-shot Modelling (PRIZM), a two-phase workflow that first uses a high-quality low-N dataset to identify the most suitable pre-trained zero-shot model for a target protein property and then applies that model to rank and prioritize an in silico variant library for experimental testing. Across diverse benchmark datasets spanning multiple protein properties, PRIZM reliably separated low- from high-performing models using datasets of [~]20 labelled variants. We further demonstrate PRIZM in enzyme engineering case studies targeting sucrose synthase thermostability and glycosyltransferase activity, where PRIZM-guided selection identified improved variants, including gains of [~]3{degrees}C in apparent melting temperature and [~]20% higher relative activity. PRIZM provides an accessible, data-efficient route to leverage foundation models for protein design while requiring minimal experimental data.

bioinformatics↗

bioRxiv2026

Investigating the Conformational Flexibility of Staphylokinase Across Multiple Time Scales

Cardiovascular diseases, including ischemic stroke, necessitate improved thrombolytic agents. A microbe-encoded plasminogen activator staphylokinase (SAK) is a promising alternative to the widely used tissue plasminogen activator (tPA) due to its high fibrin specificity and low production cost. To overcome potential immunogenicity hampering its use in clinical settings, the low-immunogenic variants SAK SY155 and SAK THR174 were previously engineered. However, the molecular basis underlying their reduced immunogenicity is not understood and requires detailed elucidation. Here, we determine molecular structures and compare flexibility between low-immunogenic and immunogenic SAK variants, using a combination of experimental and computational structural techniques. Our analyses show that all variants share the canonical SAK fold and retain similar plasminogen activation kinetics, despite the number of introduced substitutions. Crucially, the low-immunogenic variants exhibit distinct flexibility profiles, with SAK THR174 showing substantially increased flexibility in the H1 helix and B3 region. SAK SY155 exhibits an increased flexibility in the H1-B3 loop and propensity to homodimerize. These flexibility changes are found in the known immunogenic epitopes. Our multi-scale flexibility analysis provides the molecular explanation for the reduced immunogenicity, altered thermostability, and retained fibrinolytic function of the engineered variants. This information is critical for the design of next-generation thrombolytics.

biochemistry↗

bioRxiv2025

SoluProtMut: Siamese Deep Learning for Predicting Solubility Effects of Protein Mutations with Experimental Validation

Protein solubility is an attractive engineering target because it is a critical property influencing the scalability of protein production and the success of therapeutic proteins in biomedical applications. However, predicting solubility changes upon mutation in silico is challenging due to data heterogeneity and protein bias. Here, we explore how different sources of solubility data can be used for machine learning and present SoluProtMut, a Siamese deep geometric neural network trained to predict the impact of mutations on protein solubility. Our final model was trained exclusively on deep mutational scanning data. We compare our model with five established solubility prediction methods. The model achieves state-of-the-art performance on an independent dataset of various proteins, especially in predicting the effects of multipoint mutations (informedness of 26.5 %). Our findings also reaffirm that the scarcity of solubility data continues to hamper progress in this field. To address this limitation, we experimentally quantified solubility changes for hundreds of single-point and multipoint mutants of haloalkane dehalogenase. Complemented with recent deep-mutational-scanning data on myoglobin, we employed both these data for external validation. Although the generalization to unseen proteins remains limited, our findings demonstrate the potential of integrating high-throughput assays with deep learning to improve the accuracy and scope of solubility prediction. Highlights- We present a novel anti-symmetric Siamese architecture for mutational prediction on structures built on graph-based convolutional neural networks ensuring SE(3) invariance. - We demonstrate that a model trained only on single-point mutants of a single protein derived from high-throughput experiments generalizes to multipoint mutants of unseen proteins, achieving the state-of-the-art binary prediction informedness of 26.5 %. - We address the key limitation in the domain by extending the available data with 277 single-point and multipoint mutants of haloalkane dehalogenase labelled in-house and a selection of recently published 1037 single-point mutants of myoglobin. - We systematically study how different data subsets affect the performance of the trained models, revealing that including yeast-derived high-throughput data in training hampers generalization to low-throughput assays but recovers the performance on the yeast-derived myoglobin dataset.

bioinformatics↗

bioRxiv2025

Taurine Inhibits Apolipoprotein E4 Aggregation

Apolipoprotein E4 (ApoE4) is a major genetic risk factor in many neurodegenerative diseases, yet effective therapeutic strategies targeting its associated pathologies remain unresolved. The aggregation of ApoE4, a key pathological feature, can be attenuated by tramiprosate and its metabolite 3-sulfopropanoic acid. In this study, we investigated the potential of taurine, a close chemical analogue of tramiprosate, to modulate ApoE4-mediated pathological processes. Using an integrated approach--including molecular dynamics simulations, static light scattering, mass spectrometry, and cerebral organoid models--we investigated taurines effects on ApoE4 aggregation. We found that taurine effectively inhibits ApoE4 aggregation. Notably, taurine significantly ameliorates the pathophysiological characteristics of ApoE4, bringing its phenotype closer to the more benign ApoE3 variant. By leveraging its neuroprotective properties, taurine may offer effects comparable to tramiprosate and 3-sulfopropanoic acid, positioning it as an accessible and promising candidate for mitigating neurodegeneration, particularly in individuals with the high-risk ApoE4/E4 genotype.

biochemistry↗

bioRxiv2024

Analysis of Mutations in Precision Oncology using The Automated, Accurate, and User-Friendly Web Tool PredictONCO

Next-generation sequencing technology has created many new opportunities for clinical diagnostics, but it faces the challenge of functional annotation of identified mutations. Various algorithms have been developed to predict the impact of missense variants that influence oncogenic drivers. However, computational pipelines that handle biological data must integrate multiple software tools, which can add complexity and hinder non-specialist users from accessing the pipeline. Here, we have developed an online user-friendly web server tool PredictONCO that is fully automated and has a low barrier to access. The tool models the structure of the mutant protein in the first step. Next, it calculates the protein stability change, pocket level information, evolutionary conservation, and changes in ionisation of catalytic amino acid residues, and uses them as the features in the machine-learning predictor. The XGBoost-based predictor was validated on an independent subset of held-out data, demonstrating areas under the receiver operating characteristic curve (ROC) of 0.95 and 0.94, and the average precision from the precision-recall curve 0.98 and 0.94 for structure-based and sequence-based predictions, respectively. Finally, PredictONCO calculates the docking results of small molecules approved by regulatory authorities. We demonstrate the applicability of the tool by presenting its usage for variants in two cancer-associated proteins, cellular tumour antigen p53 and fibroblast growth factor receptor FGFR1. Our free web tool will assist with the interpretation of data from next-generation sequencing and navigate treatment strategies in clinical oncology: https://loschmidt.chemi.muni.cz/predictonco/.

bioinformatics↗

bioRxiv2024

A Computational Workflow for Analysis of Missense Mutations in Precision Oncology

Every year, more than 19 million cancer cases are diagnosed, and this number continues to increase annually. Since standard treatment options have varying success rates for different types of cancer, understanding the biology of an individuals tumour becomes crucial, especially for cases that are difficult to treat. Personalised high-throughput profiling, using next-generation sequencing, allows for a comprehensive examination of biopsy specimens. Furthermore, the widespread use of this technology has generated a wealth of information on cancer-specific gene alterations. However, there exists a significant gap between identified alterations and their proven impact on protein function. Here, we present a bioinformatics pipeline that enables fast analysis of a missense mutations effect on stability and function in known oncogenic proteins. This pipeline is coupled with a predictor that summarises the outputs of different tools used throughout the pipeline, providing a single probability score, achieving the balanced accuracy above 86%. The pipeline incorporates a virtual screening method to suggest potential FDA/EMA-approved drugs to be considered for treatment. We showcase three case studies to demonstrate the timely utility of this pipeline. To facilitate access and analysis of cancer-related mutations, we have packaged the pipeline as a web server, which is freely available at https://loschmidt.chemi.muni.cz/predictonco/.

cancer biology↗

bioRxiv2023

Large-scale Annotation of Biochemically Relevant Pockets and Tunnels in Cognate Enzyme-Ligand Complexes

Tunnels in enzymes with buried active sites are key structural features allowing the entry of substrates and the release of products, thus contributing to the catalytic efficiency. Targeting the bottlenecks of protein tunnels is also a powerful protein engineering strategy. However, the identification of functional tunnels in multiple protein structures is a non-trivial task that can only be addressed computationally. We present a pipeline integrating automated structural analysis with an in-house machine-learning predictor for the annotation of protein pockets, followed by the calculation of the energetics of ligand transport via biochemically relevant tunnels. A thorough validation using eight distinct molecular systems revealed that CaverDock analysis of ligand un/binding is on par with time-consuming molecular dynamics simulations, but much faster. The optimized and validated pipeline was applied to annotate more than 17,000 cognate enzyme-ligand complexes. Analysis of ligand un/binding energetics indicates that the top priority tunnel has the most favourable energies in 75 % of cases. Moreover, energy profiles of cognate ligands revealed that a simple geometry analysis can correctly identify tunnel bottlenecks only in 50 % of cases. Our study provides essential information for the interpretation of results from tunnel calculation and energy profiling in mechanistic enzymology and protein engineering. We formulated several simple rules allowing identification of biochemically relevant tunnels based on the binding pockets, tunnel geometry, and ligand transport energy profiles.

bioinformatics↗

bioRxiv2023

Effects of Alzheimer's Disease Drug Candidates on Disordered Aβ42 Dissected by Comparative Markov State Analysis (CoVAMPnet)

Computational study of the effect of drug candidates on intrinsically disordered biomolecules is challenging due to their vast and complex conformational space. Here we developed a Comparative Markov State Analysis (CoVAMPnet) framework to quantify changes in the conformational distribution and dynamics of a disordered biomolecule in the presence and absence of small organic drug candidate molecules. First, molecular dynamics trajectories are generated using enhanced sampling, in the presence and absence of small molecule drug candidates, and ensembles of soft Markov state models (MSMs) are learned for each system using unsupervised machine learning. Second, these ensembles of learned MSMs are aligned across different systems based on a solution to an optimal transport problem. Third, the directional importance of inter-residue distances for the assignment to different conformational states is assessed by a discriminative analysis of aggregated neural network gradients. This final step provides interpretability and biophysical context to the learned MSMs. We applied this novel computational framework to assess the effects of ongoing phase 3 therapeutics tramiprosate (TMP) and its metabolite 3-sulfopropanoic acid (SPA) on the disordered A{beta}42 peptide involved in Alzheimers disease. Based on adaptive sampling molecular dynamics and CoVAMPnet analysis, we observed that both TMP and SPA preserved more structured conformations of A{beta}42 by interacting non-specifically with charged residues. SPA impacted A{beta}42 more than TMP, protecting -helices and suppressing the formation of aggregation-prone {beta}-strands. Experimental biophysical analyses showed only mild effects of TMP/SPA on A{beta}42, and activity enhancement by the endogenous metabolization of TMP into SPA. Our data suggest that TMP/SPA may also target other biomolecules than A{beta} peptides. The CoVAMPnet method is broadly applicable to study the effects of drug candidates on the conformational behavior of intrinsically disordered biomolecules. TOC Graphic O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=109 SRC="FIGDIR/small/523007v2_ufig1.gif" ALT="Figure 1"> View larger version (46K): org.highwire.dtl.DTLVardef@13eea16org.highwire.dtl.DTLVardef@17a6bd1org.highwire.dtl.DTLVardef@3c6b33org.highwire.dtl.DTLVardef@a20444_HPS_FORMAT_FIGEXP M_FIG C_FIG

biophysics↗

bioRxiv2017

Computer-Assisted Engineering of Hyperstable Fibroblast Growth Factor 2

Fibroblast growth factors (FGFs) play numerous regulatory functions in complex organisms, and their corresponding therapeutic potential is of growing interest to academics and industrial researchers alike. However, applications of these proteins are limited due to their low stability in vivo and in vitro. Here we tackle this problem using a generalizable computer-assisted protein engineering strategy to create a unique modified FGF2 with nine mutations displaying unprecedented stability and uncompromised biological function.

bioengineering↗