Assessment of the success rate of two years of large-scale exome sequencing efforts to identify genes for Mendelian conditions at the University of Washington Center for Mendelian Genomics. J. X. Chong1, J. Shendure2, D. A. Nickerson2, M. J. Bamshad1,2,3, University of Washington Center for Mendelian Genomics 1) Department of Pediatrics, University of Washington, Seattle, WA; 2) Department of Genome Sciences, University of Washington, Seattle, WA; 3) Seattle Children's Hospital, Seattle, WA.
To evaluate clinical services and large-scale gene discovery using exome sequencing (ES), it is important to apply objective metrics to assess the success of finding causal genes for Mendelian conditions (i.e. solve rates). To date, reported solve rates are hard to interpret and use for comparisons across different contexts (i.e. clinical service vs. research). Because of its simplicity, solve rate has been reported as the proportion of families in which a causal variant for a Mendelian condition is identified or alternatively as the proportion of phenotypes for which the gene is identified. The former definition isnt particularly useful on its own as one could have a high solve rate by only sequencing families diagnosed with disorders with a single known gene, while the latter has multiple interpretations because the same phenotype can be caused by variants in multiple genes, variants in a single gene can cause multiple phenotypes, and causal variants often cannot be identified in all families. We developed three complementary metrics for solve rate and applied them to 600 families and 200 phenotypes studied at the University of Washington Center for Mendelian Genomics using strict criteria for variant causality and clear definitions of phenotype and gene novelty. The overall diagnostic rate, defined as the proportion of families for which a causal variant was identified, is 40%, while the diagnostic rate for the subset of families with causal variants in a gene previously known to cause their condition is 23%. This is comparable to diagnostic rates achieved by clinical ES even though the conditions we studied are biased against phenotypes expected to be explained by a known Mendelian disease gene. The gene identification rate, defined as the ratio of causal gene identifications to phenotypes, ranged from 0.56 (dominant) to 1 (X-linked) depending on inheritance model; for comparison, if a causal gene was identified for every phenotype, this ratio would approach its maximum value of one. Gene identifications within consanguineous families were frequently complicated by locus heterogeneity consistent with a lower than anticipated solve rate (0.81). Lastly, the novel discovery rate, or proportion of gene identifications in which the gene was newly discovered to underlie a Mendelian condition or the condition itself was novel/unexplained was 49%. Solve rates were further used to determine aspects of experimental design shared by successful projects.
You may contact the first author (during and after the meeting) at