Below we introduce the specific objectives that lead to the overall objectives of Recon4IMD, which are:
Classification of patients according to specific causative genetic defects in metabolic pathways leading to accelerated diagnosis
Classification of patients according to molecular mechanistic features that associate with clinical phenotypic severity, enabling personalised disease management
With respect to IMDs, the diagnostic challenge begins with a single symptomatic patient and ends with the identification of a causative genetic defect, while the stratification challenge begins with a set of patients with the same causative genetic defect and ends exploitation of compensatory or aggravating mechanisms in each patient. We will focus stratification efforts on an established cohort of patients with Gaucher disease (GD). In contrast, diagnosis must consider a broad range of candidate IMDs, so we will train and test computational models to diagnose with patient data from a broad range of IMDs and controls. Consequently, our first objective is:
Diagnosis and stratification will both be approached by a shared triad of mechanistic computational modelling approaches to genetics, enzyme structure, and metabolic networks. Specifically, we will develop and deploy software via the following objectives, each measurable by the number of patients classified:
Whole genomes will be sequenced, and enzyme structure can be predicted from most amino acid sequences. However, metabolic network modelling relies on having a computational representation of the metabolic pathways relevant for each IMD being modelled. Recon3 is a genome-scale reconstruction of a human metabolic network, containing 13,543 metabolic reactions between 4,140 unique metabolites, representing the catalytic function of 3,288 open reading frames. Due to a lack of comprehensiveness, Recon3 can currently be used to model only 633 of the ~1,450 IMDs, as either the genetic association is missing from the reconstruction, or the corresponding metabolic reaction is missing in the network for the other IMDs. Therefore, to broaden the applicability of metabolic network modelling of IMDs, the following is a strongly resourced objective:
Gaucher disease is considered a leading example in the IMD field, in terms of early diagnosis, aetiopathogenic understanding, and availability of therapies. IMDs share a set of general therapeutic approaches, but even where there are therapeutic options, there is an unmet need with respect to personalised patient management. Therefore, our objective is:
We recognise that the broader clinical impact of Recon4IMD is dependent on pursuing a variety of approaches to deploy data-driven personalised modelling software across Europe and beyond. Therefore, we will work towards regulatory approval of computer-aided patient classification strategies and work to promote the sustainability of project outcomes by:
1. Clinical recruitment and enrolment, using patients in established registries and cohorts, as well as new patients recruited by clinical IMD experts, in collaboration with patient associations. Success at clinical recruitment is measurable by the number and IMD diversity of enrolled patients relative to targets.
3. Enzyme structure-guided classification, given amino acid sequences for enzymes with variants of unknown significance (VUS), using structural bioinformatics and biophysical characterisation.
2. Metabolic network-based classification, using personalised whole-body metabolic network models driven by patient-derived physiological, metabolomic, and proteomic data.
4. Genomic classification, given a structured health record plus genomic analysis, using established statistical genetics pipelines.
5. Reconstruction of human metabolic networks, especially for missing genes and reactions corresponding to IMDs, with a systematic computational and literature-based effort to enhance the representation of lipid metabolism, including membrane compartments and membrane composition. Measurable by unique metabolites, reactions, and genes added to Recon3.
6. Personalised disease modelling to predict optimal treatment modalities, which are compared with systems level responses of in vitro disease models and the metabolomic profiles of patients on different treatment approaches. We will focus on stratification-based patient management superior to the standard-of-care for an established cohort of GD patients, measurable by patients stratified and patient outcomes.
7. Regulated software development of the clinically most promising academic pipelines, toward submission for approval for routine use in a clinical environment as a software medical device. Measurable by software submissions to regulatory authorities.
8. Exploitation via assessment of stakeholder perspectives on the exploitation of novel diagnostic and patient-management technologies and development of a plan for a European foundation to aid personalised diagnosis and management of IMDs.