By lack of functional evidence, genome-based diagnostic rates cap at approximately 50% across diverse Mendelian diseases. Here we demonstrate the effectiveness of combining genomics, transcriptomics, and, for the first time, proteomics and phenotypic descriptors, in a systematic diagnostic approach to discover the genetic cause of mitochondrial diseases. On fibroblast cell lines from 145 individuals, tandem mass tag labelled proteomics detected approximately 8,000 proteins per sample and covered over 50% of all Mendelian disease-associated genes. By providing independent functional evidence, aberrant protein expression analysis allowed validation of candidate protein-destabilising variants and of variants leading to aberrant RNA expression. Overall, our integrative computational workflow led to genetic resolution for 21% of 121 genetically unsolved cases and to the discovery of two novel disease genes. With increasing democratization of high-throughput omics assays, our approach and code provide a blueprint for implementing multi-omics based Mendelian disease diagnostics in routine clinical practice.

BackgroundThe spectrum of mitochondrial disease is genetically and phenotypically diverse, resulting from pathogenic variants in over 400 genes, with aerobic energy metabolism defects as a common denominator. Such heterogeneity poses a significant challenge in making an accurate diagnosis, critical for precision medicine.MethodsIn an international collaboration initiated by the European Network for Mitochondrial Diseases (GENOMIT) we recruited 2,023 pediatric patients at 11 specialist referral centers between October 2010 and January 2021, accumulating exome sequencing and HPO-encoded phenotype data. An exome-wide search for variants in known and potential novel disease genes, complemented by functional studies, followed ACMG guidelines.Results1,109 cases (55%) received a molecular diagnosis, of which one fifth have potential disease-modifying treatments (236/1,109, 21%). Functional studies enabled diagnostic uplift from 36% to 55% and discovery of 62 novel disease genes. Pathogenic variants were identified within genes encoding mitochondrial proteins or RNAs in 801 cases (72%), while, given extensive phenotype overlap, the remainder involved proteins targeted to other cellular compartments. To delineate genotype-phenotype associations, our data was complemented with registry and literature data to develop “GENOMITexplorer”, an open access resource detailing patient- (n=3,940), gene- (n=427), and variant-level (n=1,492) associations (prokischlab.github.io/GENOMITexplorer/).ConclusionsReaching a molecular diagnosis was essential for implementation of precision medicine and clinical trial eligibility, underlining the need for genome-wide screening given inability to accurately define mitochondrial diseases clinically. Key to diagnostic success were functional studies, encouraging early acquisition of patient- derived tissues and routine integration of high-throughput functional data to improve patient care by uplifting diagnostic rate.

Over the last 5 years, RNA sequencing (RNA‐seq) has been established and is increasingly applied as an effective approach complementary to DNA sequencing in molecular diagnostics. Currently, three RNA phenotypes, aberrant expression, aberrant splicing, and allelic imbalance, are considered to provide information about pathogenic variants. By providing a high‐throughput, transcriptome‐wide functional readout on variants causing aberrant RNA phenotypes, RNA‐seq has increased diagnostic rates by about 15% over whole‐exome sequencing. This breakthrough encouraged the development of computational tools and pipelines aiming to streamline RNA‐seq analysis for implementation in clinical diagnostics. Although a number of studies showed the added value of RNA‐seq for the molecular diagnosis of individuals with Mendelian disorders, there is no formal consensus on assessing variant pathogenicity strength based on RNA phenotypes. Taking RNA‐seq as a functional assay for genetic variants, we evaluated the value of statistical significance and effect size of RNA phenotypes as evidence for the strength of variant pathogenicity. This was determined by the analysis of 394 pathogenic variants, of which 198 were associated with aberrant RNA phenotypes and 723 benign variants. Overall, this study seeks to establish recommendations for integrating functional RNA‐seq data into the the American College of Medical Genetics and Genomics and the Association for Molecular Pathology guidelines classification system.