Tomáš Helikar

SUMMARY: Identifying robust biomarkers from high-dimensional biomedical data is a central challenge in translational research, but candidate rankings produced by any single feature-selection or classification method depend on algorithmic choices and rarely reproduce across pipelines. We present a disease-agnostic machine-learning framework that addresses this dependence by systematically benchmarking 25 (feature-selection x classifier) pipelines under five-fold stratified cross-validation, aggregating per-feature evidence by two independent methods (a weighted-selection consensus score and Robust Rank Aggregation), and characterizing the direction of each candidate using Cohen’s d . We demonstrate the framework on immune-response measurements from two clinical phases: SARS-CoV-2 hospitalization and intensive-care admission; obtaining cross-validated mean F1 above 0.99 with balanced classification errors and producing tiered, direction-aware biomarker lists per phase. Interleukin-18 (IL-18) reached the strongest tier in both phases with consistent direction. The framework generalizes to any binary clinical classification problem and supports principled, reproducible biomarker prioritization.

Logical models are widely used to study regulatory and signaling systems, yet their reuse, annotation, and exchange across tools remain challenging. Although SBML Level 3 Qualitative Models (SBML-qual) provides a standard representation, its XML-based syntax is difficult to inspect and edit directly. Here we introduce TabularQual, a spreadsheet-based, community-driven representation for Boolean and multi-valued logical models, together with a bidirectional converter between spreadsheets and SBML-qual. The converter is accessible programmatically and via a web interface to support diverse user workflows. We further describe two integration strategies that enable existing modeling tools to operate with TabularQual, either through SBML-qual exchange or via direct support. Case studies using the Stress Knowledge Map and CaSQ demonstrate how this integration supports model construction, curation, and reuse. TabularQual provides a practical bridge between human-readable model representations and standardized executable formats, supporting reproducibility, interoperability, and community-driven model development.

Constraint-based metabolic modeling is a powerful way to study the mechanistic basis of cellular states and disease, but its effective use demands substantial computational expertise and careful coordination of multi-step analyses. We developed MechAInistic to lower this barrier and enable researchers to ask complex biological questions in natural language. Harnessing large language models, MechAInistic is a multi-agent system organized around an Architect-Reviewer pattern that transforms a natural-language question into an executable, model-grounded workflow and generates a structured report. The system supports a variety of tasks, including pathway comparison, perturbation analysis, drug-target exploration, and literature-grounded interpretation across paired metabolic model states. We tested MechAInistic on two drug-repurposing use cases. For Naive B cells from Rheumatoid Arthritis (RA) paired with healthy controls, the system quantified the metabolic rewiring driving disease, prioritized candidate reactions using topological hub filtering and robustness analysis, and surfaced Devimistat as a potential repurposing candidate acting through 2-oxoglutarate dehydrogenase in the TCA cycle. In a paired CD4+ Th17 cell study from Multiple Sclerosis (MS) and healthy controls, the same workflow identified NADP-dependent isocitrate dehydrogenase as the optimal single target and proposed ivosidenib as an FDA-approved repurposing candidate. Together, these results show that MechAInistic interfaces directly with mechanistic modeling and turns large language model reasoning into reproducible biological discovery. MechAInistic is accessible at https://mechainistic.dtih.org.

Introduction: The immune system is a dynamic, complex network of interacting cells, molecules, and signals central to health. Immune digital twins - virtual representations of the immune system that evolve in tandem with their biological counterparts - offer a path toward predictive, patient-specific simulations, but their realization requires mechanistic frameworks that generalize across multiple cell types and disease contexts. Methods: We constructed a logic-based mechanistic model of the human immune system built exclusively on human experimental data extracted from 449 publications. The model integrates 51 innate (e.g., NK cells, macrophages) and 37 adaptive (e.g., Th1/2/17, B cells) immune cell types and subtypes, together with 37 secretory factors and 1,450 regulatory interactions across 11 disease conditions, including nine pathogens, type 1 diabetes, and lung transplantation. Model predictions were validated against independent in vitro, ex vivo, and clinical observations not used during construction. Results: Across 38 validation experiments, agreement with published literature ranged from 75% to 90% across pathogens. The model reproduced pathogen-specific cytokine signatures for nine infections, captured synergistic and antagonistic effects in four coinfection scenarios (MTB–HIV, MTB–Helminth, SARS-CoV-2–EBV, and Plasmodium falciparum -Helminth), and resolved competing rejection and tolerance signals in lung transplantation challenged by CMV, EBV, and SARS-CoV-2. Discussion: This reference model unifies multiple immune contexts within a single simulatable model and generates a structured catalog of falsifiable predictions for experimental follow-up. The model is openly available on Cell Collective and as an SBML file, representing a step toward a clinically integrated immune digital twin for patient-specific applications.

Summary: Improved accessibility of high-throughput RNA sequencing has increased the amount of data generated each year. This increase in data creates a need for reproducible pipelines that can process RNA-seq data consistently across experiments. AutoRNAseq addresses this need by providing a Snakemake-based workflow for bulk RNA-seq analysis by automating data retrieval, quality control, and gene quantification. Unlike existing RNA-seq workflows that require users to coordinate multiple pipelines and pre-configure reference data, AutoRNAseq provides a single, end-to-end workflow that automates data acquisition, reference preparation, quality control, alignment, and quantification with minimal user intervention. AutoRNAseq is applicable to any domain requiring consistently processed RNA-seq datasets, including bioinformatics, computational biology, and drug-response studies. Availability and Implementation: AutoRNAseq is implemented in Snakemake and available at https://gitlab.com/unebraska/lagbh-public/autornaseq . Documentation and example configuration files are provided in the GitLab README file and this paper’s Supplementary Information. The code to reproduce the statistics presented here is in the GitLab repository under the “publication” folder.

Acute radiation syndrome (ARS) occurs when hematopoietic or gastrointestinal cells are damaged by radiation exposure causing DNA damage to the bone marrow and gastrointestinal epithelial stem cell populations. In these highly proliferative cell types, DNA damage inhibits stem cell repopulation. In humans and animals, this inability to regenerate stem cells is lethal. Within this manuscript, several compounds, Amifostine, Captopril, Ciprofloxacin, PrC-210, 5-AED (5-androstene-3β,17β-diol), and 5-AET (5-androstene-3β,7β,17B-triol), are assessed for their ability to protect against ARS in an in vitro and/or in vivo setting. ARS was accomplished by irradiating mouse bone marrow cells or rat intestinal epithelial (IEC-6) cells in vitro with 4–8 Gy and in vivo by exposing Mus musculus to 7.3 Gy of whole-body irradiation. The primary endpoints of this study include cellular viability, DNA damage via γ-H2AX, colony formation, and overall survival at 30-days post-irradiation. In addition to evaluating the radioprotective performance of each compound, this study establishes a distinct set of in vitro assays to predict the overall efficacy of potential radioprotectors in an in vivo model of ARS. Furthermore, these results highlight the need for FDA-approved medical intervention to protect against ARS.

Artificial intelligence (AI) is driving innovation in clinical pharmacology and translational science with tools to advance drug development, clinical trials, and patient care. This review summarizes the key takeaways from the AI preconference at the American Society for Clinical Pharmacology and Therapeutics (ASCPT) 2024 Annual Meeting in Colorado Springs, where experts from academia, industry, and regulatory bodies discussed how AI is streamlining drug discovery, dosing strategies, outcome assessment, and patient care. The theme of the preconference was centered around how AI can empower clinical pharmacologists and translational researchers to make informed decisions and translate research findings into practice. The preconference also looked at the impact of large language models in biomedical research and how these tools are democratizing data analysis and empowering researchers. The application of explainable AI in predicting drug efficacy and safety, and the ethical considerations that should be applied when integrating AI into clinical and biomedical research were also touched upon. By sharing these diverse perspectives and real‐world examples, this review shows how AI can be used in clinical pharmacology and translational science to bring efficiency and accelerate drug discovery and development to address patients' unmet clinical needs.

Guidelines for managing scientific data have been established under the FAIR principles requiring that data be Findable, Accessible, Interoperable, and Reusable. In many scientific disciplines, especially computational biology, both data and models are key to progress. For this reason, and recognizing that such models are a very special type of 'data', we argue that computational models, especially mechanistic models prevalent in medicine, physiology and systems biology, deserve a complementary set of guidelines. We propose the CURE principles, emphasizing that models should be Credible, Understandable, Reproducible, and Extensible. We delve into each principle, discussing verification, validation, and uncertainty quantification for model credibility; the clarity of model descriptions and annotations for understandability; adherence to standards and open science practices for reproducibility; and the use of open standards and modular code for extensibility and reuse. We outline recommended and baseline requirements for each aspect of CURE, aiming to enhance the impact and trustworthiness of computational models, particularly in biomedical applications where credibility is paramount. Our perspective underscores the need for a more disciplined approach to modeling, aligning with emerging trends such as Digital Twins and emphasizing the importance of data and modeling standards for interoperability and reuse. Finally, we emphasize that given the non-trivial effort required to implement the guidelines, the community moves to automate as many of the guidelines as possible.

Genome-scale metabolic models (GEMs) and other constraint-based models (CBMs) play a pivotal role in understanding biological phenotypes and advancing research in areas like metabolic engineering, human disease modelling, drug discovery, and personalized medicine. Despite their growing application, a significant challenge remains in ensuring the reproducibility of GEMs, primarily due to inconsistent reporting and inadequate model documentation of model results. Addressing this gap, we introduce FROG analysis, a community-driven initiative aimed at standardizing reproducibility assessments of CBMs and GEMs. The FROG framework encompasses four key analyses—Flux variability, Reaction deletion, Objective function, and Gene deletion—to produce standardized, numerically reproducible FROG reports. These reports serve as reference datasets, enabling model evaluators, curators, and independent researchers to verify the reproducibility of GEMs systematically. BioModels, a leading repository of systems biology models, has integrated FROG analysis into its curation workflow, enhancing the reproducibility and reusability of submitted GEMs. In our study evaluating 65 GEM submissions from the community, approximately 40% reproduced without intervention, 28% requiring minor adjustments, and 32% needing input from authors. The standardization introduced by FROG analysis facilitated the detection and resolution of issues, ultimately leading to the successful reproduction of all models. By establishing a standardized and comprehensive approach to evaluating GEM reproducibility, FROG analysis significantly contributes to making CBMs and GEMs more transparent, reusable, and reliable for the broader scientific community.

Modern biological research is increasingly informed by computational simulation experiments, which necessitate the development of methods for annotating, archiving, sharing, and reproducing the conducted experiments. These simulations increasingly require extensive collaboration among modelers, experimentalists, and engineers. The Minimum Information About a Simulation Experiment (MIASE) guidelines outline the information needed to share simulation experiments. SED-ML is a computer-readable format for the information outlined by MIASE, created as a community project and supported by many investigators and software tools. Level 1 Version 5 of SED-ML expands the ability of modelers to define simulations in SED-ML using the Kinetic Simulation Algorithm Onotoloy (KiSAO). While it was possible in Version 4 to define a simulation entirely using KiSAO, Version 5 now allows users to define tasks, model changes, ranges, and outputs using the ontology as well. SED-ML is supported by a growing ecosystem of investigators, model languages, and software tools, including various languages for constraint-based, kinetic, qualitative, rule-based, and spatial models, and many simulation tools, visual editors, model repositories, and validators. Additional information about SED-ML is available at https://sed-ml.org/.