Table 3.
ClassifyCNV (%) | AnnotSV (%) | ||
---|---|---|---|
Pathogenic/likely pathogenic | Sensitivity | 60.9 | 100 |
Specificity | 98.4 | 19.7 | |
Accuracy | 88.1 | 41.7 | |
% Benign/likely benign calls | 0 | 0 | |
% Uncertain calls | 39.1 | 0 | |
Benign/likely benign | Sensitivity | 25 | 37.5 |
Specificity | 100 | 92.1 | |
Accuracy | 92.9 | 86.9 | |
% Pathogenic/likely pathogenic calls | 0 | 62.5 | |
% Uncertain calls | 75 | 0 | |
Uncertain significance | Sensitivity | 98.1 | 5.6 |
Specificity | 51.6 | 100 | |
Accuracy | 81 | 40.5 |
Both tools were tested on a set of 84 variants manually evaluated by ACMG/ClinGen. While ClassifyCNV produced a higher percentage of uncertain calls compared to AnnotSV, it had higher specificity and accuracy for pathogenic/likely pathogenic and benign/likely benign variants.