Flexible LR-based assignment of evidence weightings, as described in the 2018 Bayesian evolution of the ACMG/AMP framework
| LR ratio range | Evidence points | Direction of evidence | ACMG/AMP (2015) evidence strength |
| ≥350.4 (2.088) | 8 | Towards pathogenicity | Very strong |
| ≥18.72 (2.084) and <350.4 | 4 | Towards pathogenicity | Strong |
| ≥4.33 (2.082) and <18.72 | 2 | Towards pathogenicity | Moderate |
| ≥2.08 (2.081) and <4.33 | 1 | Towards pathogenicity | Supporting |
| ≤0.48 (2.08−1) and >0.23 | −1 | Towards benignity | Supporting |
| ≤0.23 (2.08−2) and >0.053 | −2 | Towards benignity | Moderate |
| ≤0.053 (2.08−4) and >0.00285 | −4 | Towards benignity | Strong |
| ≤0.00285 (2.08−8) | −8 | Towards benignity | Very strong |
PS4-LRCalc was developed in Python (V.3.11), and analyses were performed using PyCharm (V.23.1.1, Professional Edition) for remote development on a high-performance computing cluster. The online tool for LR calculator use was developed using Shiny for Python.
As described by Tavtigian et al, evidence criteria for which a likelihood ratio towards pathogenicity can be quantified may be converted to EPs through log-transformation (to base 2.08).13 14 This continuous approach reflects evidence strength quantitatively, in contrast to the categorical approach of the 2015 ACMG/AMP framework.
ACMG/AMP, American College of Medical Genetics/Association of Molecular Pathology; EPs, exponent points; LR, likelihood ratio; PS4-LRCalc, PS4-likelihood ratio calculator.