AimsMitochondrial diseases are rare diseases which are phenotypically and genetically heterogeneous. About two thirds of mitochondrial diseases are caused by mitochondrial DNA (mtDNA) variants, the rest by nuclear DNA variants. To improve mtDNA variant interpretation, it is important to gather published evidence of phenotype-genotype associations.Aims areTo build a manually curated database of mitochondrial DNA (mtDNA) disease phenotypes and genotypes, using the human phenotype ontology (HPO).To use this database to identify patients with mtDNA disease in rare disease sequencing projects.MethodsWe independently reclassified mtDNA variants from MITOMAP and ClinVar through online literature review of pathogenicity criteria.The literature on each pathogenic variant was manually curated to provide individual-level phenotype-genotype data in a relational database called MitoPhen.Phenotype similarity scores were performed between probands in MitoPhen and patients with confirmed mtDNA diseases and individuals with a non-mitochondrial nuclear genetic disorder enrolled in the NIHR BioResource Rare Diseases study and the Solve-RD research project.Results89 mtDNA variants (4 indels, 85 single nucleotide variants), fulfilled criteria for pathogenicity. 676 publications were used to populate MitoPhen. We curated data from 6688 individuals, 3696 (55%) were recorded as clinically affected. 1349 (20%) are affected patients with Paediatric-onset disease. 26348 HPO terms were recorded across 3800 individuals (including 118 individuals noted as asymptomatic). The mean number of terms per proband is 11.4.Using mean phenotype similarity scores computed through MitoPhen, we were able to show that patients with mtDNA disease could be distinguished from non-mitochondrial rare diseases (including other neurodevelopmental disorders). There was no statistically significant difference between phenotype similarity scores computed for adult-onset and paediatric-onset mtDNA disease patients.ConclusionTo direct patients to appropriate genetic counselling earlier, our clinical interpretation of potentially pathogenic variants must be efficient. Current variant annotation pipelines in rare disease sequencing projects are supplemented by HPO analyses, however a reference phenotype dataset for mtDNA disease does not exist.MitoPhen, found at www.mitophen.org, provides the first manually curated database for mtDNA disease that could be used to discover mtDNA diagnoses in large sequencing projects. Further work is needed to enrich the database with rare pathogenic variants....

580 MitoPhen: A human phenotype ontology-based tool to identify mitochondrial DNA disease

10.1136/archdischild-2022-rcpch.359