A Systems Biology Workflow to Support the Diagnosis of Inherited Metabolic Disorders: a study on Pyrimidine and Urea Cycle disorders

Abstract BackgroundInherited Metabolic Disorders (IMDs) are rare diseases where one impaired protein leads to a cascade of changes in the adjacent chemical conversions. These disorders often present with non-specific symptoms, a lack of a clear genotype-phenotype correlation, and de novo mutations, complicating diagnosis. Furthermore, products of one metabolic conversion can be the substrate of another pathway (much further) down- or upstream, obscuring biomarker identification and causing overlapping biomarkers for different disorders. Last, current clinical diagnoses are lacking a visualization of the connections between individual metabolic biomarkers and the enzymes involved in their synthesis and degradation. The goal of this study was to provide a proof-of-concept systems biology framework for diagnostic support of IMDs by combining clinical and theoretical metabolic biomarkers with pathway models using network approaches and semantic web technologies. This workflow could support the diagnosis of other less understood IMDs, by building on two groups of well-studied metabolic disorders linked to the pyrimidine and urea cycle pathways. MethodsOur framework integrates literature and expert knowledge into machine-readable pathways. The top three pathways of interest were retrieved covering most unique patient markers and overlapping with theoretical marker data, which was used to visualize clinical data of 16 previously diagnosed patients with various pyrimidine and urea cycle disorders. Two expert laboratory scientists evaluated the resulting visualizations to derive a diagnosis. ResultsThe number of relevant biomarkers for each patient varied greatly (five to 48), and the pathways covering most unique biomarkers differ for equivalent disorders. The two experts reached the same conclusions for all samples with our proposed framework for nine patient samples, without knowledge about clinical symptoms or sex. For the remaining seven cases, four interpretations pointed in the direction of a subset of disorders, while three cases were found to be undiagnosable with the available data. ConclusionThe presented workflow integrates chemical reactions with clinical data in a visual representation and is adaptable to analyze different types of IMDs (known and novel), difficult patient cases, and functional assays in the future. By using semantic web technologies and network approaches, the workflow could easily be extended with other OMICS datasets, as well as linked to knowledge captured as Linked Open Data.