Table 4. Exemplar case-control rare disease-type scenarios.
| Scenario number | Case number with variant/total cases | Control number with variant/total controls | Observed effect size | Likelihood OR≥10 versus OR≤1 | Likelihood OR≥1000 versus OR≤1 | |||
| OR (95% CI) | p-value (Fisher’s exact test) | LR | EPs | LR | EPs | |||
| 12 | 2/30 | 2/300 000 | 10 700 (1458 to 78 746) | 5.80×10–8 | 1.23×1011 | 34.9 | 1.22×1011 | 34.9 |
| 13 | 5/30 | 20/300 000 | 3000 (1044 to 8619) | 3.73×10–16 | 1.30×1019 | 60.1 | 1.28×1019 | 60.1 |
| 14 | 2/300 | 20/300 000 | 101 (23 to 433) | 2.27×10–4 | 5.86×105 | 18.1 | 20.63 | 4.1 |
| 15 | 5/300 | 20/300 000 | 254 (95 to 682) | 5.03×10–11 | 4.91×1012 | 39.9 | 7.03×109 | 31.0 |
| 16 | 5/300 | 50/300 000 | 102 (40 to 257) | 3.21×10–9 | 3.57×1010 | 33.2 | 3.1 | 1.5 |
| 17 | 3/30 | 15/15 000 | 111 (30 to 406) | 5.74×10–6 | 2.52×107 | 23.3 | 497.1 | 8.5 |
| 18 | 3/30 | 30/15 000 | 55 (16 to 193) | 3.76×10–5 | 2.14×106 | 19.9 | 4.75×10–5 | −13.6 |
In the seven scenarios shown here, as may be typical in a rare-disease clinical setting, the frequency of variants in small-to-modest case series is evaluated against variant frequencies as might be derived from population-scale data sets, using appropriate ancestry-matched subsets therein. All scenarios would attain PS4 under the 2015 ACMG/AMP specification of OR>5 and p<0.05. We apply PS4-LRCalc under hypotheses of association of OR≥10 and OR≥1000, the latter arguably being more consistent with very rare disease. We illustrate in scenarios 14,16 and 18 the substantial impact of differing this target OR against which observed data are evaluated. These use cases assume fully robust genotyping and phenotyping: sensitivity assessment and adjustment may be warranted in scenarios where numbers are small and/or there is uncertainty regarding the robustness of data quality/ascertainment (eg, in scenarios 12 and 14 in which there are only two case observations). ORs and associated 95% CIs are rounded to the nearest whole number here for ease of visualisation; full data are displayed in online supplemental file 1).
ACMG/AMPAmerican College of Medical Genetics/Association of Molecular PathologyEPexponent pointLRlikelihood ratioORodds ratioPS4-LRCalcPS4-likelihood ratio calculator