Scenario number | Case number with variant/total cases | Control number with variant/total controls | Observed effect size | ACMG 2015 Framework (OR≥5, LCI≥1) | Likelihood of OR≥5 versus likelihood of OR≤1 | ||

OR (95% CI) | p-value (Fisher’s exact test) | LR | EPs | ||||

7 | 1/10 000 | 1/10 000 | 1.0 (0.06 to 15.99) | 1.00 | None | 0.15 | −2.6 |

8 | 2/10 000 | 2/10 000 | 1.0 (0.14 to 7.10) | 1.00 | None | 7.10×10^{–2}
| −3.6 |

9 | 10/10 000 | 10/10 000 | 1.0 (0.42 to 2.40) | 1.00 | None | 3.75×10^{–4}
| −10.8 |

10 | 2/10 000 | 4/10 000 | 0.5 (0.09 to 2.73) | 0.69 | None | 2.59×10^{–3}
| −8.1 |

11 | 2/10 000 | 10/10 000 | 0.2 (0.04 to 0.91) | 0.04 | None | 1.55×10^{–7}
| −21.4 |

Shown are illustrative sets of case-control observations in data sets of equivalent size and their equivalent likelihood ratios (LRs) and exponent points (EPs) under the PS4-LRCalc model. The 2015 ACMG/AMP framework does not permit the use of lack of case-control signal as evidence for benignity. However, in scenarios 7–9, calculation of LRs using PS4-LRCalc allows application of PS4 in the benign direction at increasing strength—from the equivalent of moderate to very strong—as the number of variant observations increases. In scenarios 10 and 11, EPs for variants with ostensibly protective effects, that is, observed at higher frequency in controls than cases, reach the equivalent of ‘very strong’ in the benign direction. Scenarios 8, 9 and 10 are further illustrated in figure 2D–F. LCI, lower confidence interval; OR, odds ratio.

ACMG, American College of Medical Genetics; AMP, Association of Molecular Pathology; EP, exponent point; LR, likelihood ratio; OR, odds ratio; PS4-LRCalc, PS4-likelihood ratio calculator.