Periodic Reporting for period 1 - Recon4IMD (Reconstruction and Computational Modelling for Inherited Metabolic Diseases)
Okres sprawozdawczy: 2023-06-01 do 2024-11-30
Our goal is to accelerate diagnosis and enable personalized management of inherited metabolic diseases (IMDs). We will apply advanced statistical genomics, deep learning-based protein structure prediction, and whole-body metabolic network modeling to generate patient-specific computational models using genomic, transcriptomic, proteomic, and metabolomic data. A clinical team will recruit 1,945 diagnosed IMD patients to train diagnostic models and validate them on 685 undiagnosed cases. Enhancing human metabolic network reconstruction, particularly in lipid metabolism, reaction kinetics, and IMD pathways, will improve predictive capabilities. Gaucher disease will serve as a model to identify compensatory and aggravating mechanisms affecting clinical severity. Model validation will include empirical investigations of protein structure, function, and patient-derived disease models through metabolomics, tracer-based metabolomics, and proteomics. The software developed will be broadly applicable to IMDs, accessible to clinicians, and regulatory-compliant. A roadmap for a European foundation will ensure long-term impact through broad stakeholder consultation. This initiative leverages European collaboration to advance personalized computational modeling for rare diseases.
The Recon4IMD project has made significant strides in advancing its two overarching objectives of accelerating IMD diagnosis through computational modeling and stratifying patients using personalized approaches. Across all work packages, progress reflects a concerted effort to integrate clinical data, computational tools, and experimental techniques to achieve these goals.
Clinical recruitment efforts have established a robust framework for patient enrollment and data collection, including the IMD-Hub platform and multilingual study documents, ensuring broad participation from diverse IMD cohorts. Ethical approvals, patient engagement via UNIAMO, and interoperability with the U-IMD registry have laid the groundwork for robust data collection, directly supporting both the diagnosis acceleration and stratification objectives by ensuring a diverse dataset for computational and experimental modeling.
Metabolic network-based classification has achieved milestones in metabolomic and proteomic analyses, addressing logistical challenges and enabling cross-platform compatibility. The development of the PROTRIDER algorithm and high-throughput assays has enhanced data quality, while whole-body metabolic modeling has identified critical biomarkers for 28 IMDs, integrating external modifiers for refined predictions. These advances align closely with the first objective, providing computational tools to accelerate diagnosis and predict candidate genes linked to altered metabolites.
In enzyme structure-guided classification, deep learning methods for protein structure prediction have been evaluated, with promising techniques identified for further investigation. The experimental validation of GBA variants through structural characterization and biophysical assays has progressed, contributing to a more nuanced understanding of missense mutations. These findings bridge computational predictions with experimental data, advancing both diagnosis acceleration and stratification efforts.
Genomic classification of variants of unknown significance has progressed through benchmarking genomic datasets and integrating outputs from enzyme structure and metabolic models. This work has led to improved clinical interpretations and confirmed diagnoses for new disease genes, demonstrating the practical impact of computational tools in stratifying patients and identifying clinically actionable variants, thus fulfilling the second objective.
The reconstruction of human metabolic networks has been bolstered by the development of the ReconX Knowledge Graph, facilitating the integration of lipid metabolism, organellar membranes, and sex-specific models. Enhancement of whole-body models provide a comprehensive framework for IMD-specific modelling, supporting both objectives by enabling personalized modeling and expanding the understanding of metabolic mechanisms.
Personalized disease modeling has progressed with the development of a computational framework for integrating transcriptomic, proteomic, and metabolomic data to simulate Gaucher disease-specific metabolic changes. Experimental validation using macrophage models and omics workflows is ongoing to test predictions and provide actionable insights into disease phenotypes and potential treatments. These advancements directly address the second objective by stratifying patients and predicting optimal therapeutic approaches.
Efforts in software medical device development have laid a strong foundation for translating academic tools into clinically compliant software. The Process Reference and Assessment Models align with regulatory requirements, ensuring the development of accessible and validated diagnostic tools. Collaborative efforts between academic and clinical teams, supported by MSc student projects, promise to deliver software solutions tailored to realistic clinical scenarios, ultimately enhancing the practical application of computational models for both objectives.
Collectively, these achievements demonstrate the project's significant progress in leveraging computational modeling and personalized approaches to address the challenges of IMD diagnosis and stratification. By integrating clinical data, experimental insights, and innovative software tools, Recon4IMD is advancing toward its vision of transforming IMD management.
Clinical recruitment efforts have established a robust framework for patient enrollment and data collection, including the IMD-Hub platform and multilingual study documents, ensuring broad participation from diverse IMD cohorts. Ethical approvals, patient engagement via UNIAMO, and interoperability with the U-IMD registry have laid the groundwork for robust data collection, directly supporting both the diagnosis acceleration and stratification objectives by ensuring a diverse dataset for computational and experimental modeling.
Metabolic network-based classification has achieved milestones in metabolomic and proteomic analyses, addressing logistical challenges and enabling cross-platform compatibility. The development of the PROTRIDER algorithm and high-throughput assays has enhanced data quality, while whole-body metabolic modeling has identified critical biomarkers for 28 IMDs, integrating external modifiers for refined predictions. These advances align closely with the first objective, providing computational tools to accelerate diagnosis and predict candidate genes linked to altered metabolites.
In enzyme structure-guided classification, deep learning methods for protein structure prediction have been evaluated, with promising techniques identified for further investigation. The experimental validation of GBA variants through structural characterization and biophysical assays has progressed, contributing to a more nuanced understanding of missense mutations. These findings bridge computational predictions with experimental data, advancing both diagnosis acceleration and stratification efforts.
Genomic classification of variants of unknown significance has progressed through benchmarking genomic datasets and integrating outputs from enzyme structure and metabolic models. This work has led to improved clinical interpretations and confirmed diagnoses for new disease genes, demonstrating the practical impact of computational tools in stratifying patients and identifying clinically actionable variants, thus fulfilling the second objective.
The reconstruction of human metabolic networks has been bolstered by the development of the ReconX Knowledge Graph, facilitating the integration of lipid metabolism, organellar membranes, and sex-specific models. Enhancement of whole-body models provide a comprehensive framework for IMD-specific modelling, supporting both objectives by enabling personalized modeling and expanding the understanding of metabolic mechanisms.
Personalized disease modeling has progressed with the development of a computational framework for integrating transcriptomic, proteomic, and metabolomic data to simulate Gaucher disease-specific metabolic changes. Experimental validation using macrophage models and omics workflows is ongoing to test predictions and provide actionable insights into disease phenotypes and potential treatments. These advancements directly address the second objective by stratifying patients and predicting optimal therapeutic approaches.
Efforts in software medical device development have laid a strong foundation for translating academic tools into clinically compliant software. The Process Reference and Assessment Models align with regulatory requirements, ensuring the development of accessible and validated diagnostic tools. Collaborative efforts between academic and clinical teams, supported by MSc student projects, promise to deliver software solutions tailored to realistic clinical scenarios, ultimately enhancing the practical application of computational models for both objectives.
Collectively, these achievements demonstrate the project's significant progress in leveraging computational modeling and personalized approaches to address the challenges of IMD diagnosis and stratification. By integrating clinical data, experimental insights, and innovative software tools, Recon4IMD is advancing toward its vision of transforming IMD management.
There is no result for this project yet