Skip to main content

Advertisement

Table 2 Refinement of etiological fractions for 6 HCM genes using variant clustering and functional prediction scores

From: Quantitative approaches to variant classification increase the yield and precision of genetic testing in Mendelian diseases: the case of hypertrophic cardiomyopathy

Gene Case excess EF (whole gene) Predictor method Prioritised variants Variants not prioritised
Case freq. EF Case freq. EF
MYH7 12.76% 0.930 (0.923–0.938) HCM cluster 10.70% 0.976 (0.972–0.981) 3.17% 0.746 (0.706–0.785)
Consensus 12.55% 0.940 (0.933–0.947) 1.32% 0.783 (0.728–0.839)
MetaSVM 12.53% 0.944 (0.937–0.951) 1.34% 0.739 (0.675–0.804)
MetaLR 13.29% 0.944 (0.938–0.951) 0.58% (p = 0.0155) 0.406 (0.185–0.627)
MYBPC3 7.98% 0.879 (0.865–0.893) HCM cluster 2.80% 0.979 (0.971–0.987) 6.39% 0.830 (0.809–0.850)
Consensus 8.42% 0.904 (0.892–0.916) 0.77% 0.524 (0.379–0.670)
MetaSVM 4.27% 0.945 (0.934–0.957) 4.92% 0.811 (0.786–0.837)
MetaLR 1.78% 0.900 (0.874–0.925) 7.41% 0.871 (0.855–0.887)
TNNT2 1.54% 0.912 (0.889–0.935) HCM cluster 1.23% 0.958 (0.941–0.974) 0.46% 0.787 (0.699–0.874)
Consensus 1.20% 0.909 (0.880–0.937) 0.49% 0.832 (0.730–0.934)
MetaSVM 1.11% 0.894 (0.861–0.927) 0.58% 0.905 (0.848–0.961)
MetaLR 1.11% 0.889 (0.856–0.923) 0.58% 0.921 (0.872–0.971)
TNNI3 1.95% 0.930 (0.912–0.948) HCM cluster 1.92% 0.974 (0.963–0.984) 0.18% (p = 0.0918) 0.457 (0.140–0.774)
Consensus 1.93% 0.957 (0.943–0.970) 0.17% (p = 0.0383) 0.566 (0.280–0.852)
MetaSVM 1.77% 0.939 (0.921–0.957) 0.33% 0.873 (0.803–0.944)
MetaLR 1.87% 0.932 (0.913–0.951) 0.23% 0.903 (0.833–0.973)
MYL3 0.70% 0.833 (0.772–0.895) HCM cluster 0.55% 0.925 (0.886–0.965) 0.29% (p = 0.0021) 0.655 (0.455–0.856)
Consensus 0.79% 0.869 (0.817–0.921) 0.05% (p = 0.6503) 0.310 (0–1)
MetaSVM 0.50% 0.840 (0.763–0.917) 0.34% 0.833 (0.735–0.930)
MetaLR 0.53% 0.809 (0.722–0.897) 0.31% 0.883 (0.809–0.958)
CSRP3 0.41% 0.683 (0.563–0.803) HCM cluster 0.43% 0.882 (0.821–0.943) 0.16% (p = 0.5533) 0.158 (0–0.724)
Consensus 0.58% 0.735 (0.630–0.839) 0.02% (p = 1.0000)
MetaSVM 0.53% 0.779 (0.687–0.871) 0.07% (p = 1.0000)
MetaLR 0.55% 0.751 (0.651–0.852) 0.05% (p = 1.0000)
  1. Comparison of performance of variant clustering and consensus functional prediction scores in enriching for disease-associated non-truncating/missense variants in 6 HCM genes where the clustering of case variants was detected. For each gene, the EF of all rare variants is shown, followed by the EF of variants prioritised by the approach, and the EF of the remaining variants that are not prioritised. Clustering analyses identified regions of 4 genes with an EF ≥ 0.95 (bold), and generally outperformed consensus functional prediction scores. Fisher’s exact p values for comparison of rare variation in cases and ExAC reference samples were < 0.0001 unless otherwise noted. For MYBPC3 (italics), the FATHMM predictor was not included in the consensus scores due to its poor performance for this gene, which also affected the MetaSVM and MetaLR consensus scores