Fig 1.
Overview of ACMG/AMP evidence-based criteria.
Green marks show implemented criteria. Grey font shows not implemented criteria. Striked-through shows removed or not applicable criterion for hearing loss. Thresholds shown for BA1 and BS1 are specific for HL.
Table 1.
Overview of ACMG classification tools benchmarked against GenOtoScope.
Fig 2.
Conceptual workflow of GenOtoScope.
Fig 3.
(A) The home page of the GenOtoScope website. (B) The resulted variant classification page, for an example variant (RS id: 1064797096), which includes its classification based on HL-specified ACMG guidelines.
Fig 4.
Command line examples for the two commands of GenOtoScope.
(A) Annotate all variants presented in VCF files, in input folder, using megSAP application and save results in GSvar files. (B) Classify all variants presented in GSvar files based on ACMG guidelines specified for HL.
Fig 5.
Conceptual flowchart to assess NMD for the refined PVS1 rule.
Fig 6.
Conceptual flowchart for examining PS1 and PM5 (PM5 Strong).
Fig 7.
ROC curves and AUC scores of all classification tools for VCEP-HL data set.
(A) Prediction of “Benign” broader class versus “Pathogenic” broader class and “VUS” class (B) Prediction of “Pathogenic” broader class versus “Benign” broader class and “VUS” class (C) Prediction “VUS” class versus “Benign” broader class and “Pathogenic” broader class.
Fig 8.
Precision-recall curves and AUC scores of all classification tools for VCEP-HL data set.
(A) Prediction of “Benign” broader class versus “Pathogenic” broader class and “VUS” class (B) Prediction of “Pathogenic” broader class versus “Benign” broader class and “VUS” class (C) Prediction of “VUS” class versus “Benign” broader class and “Pathogenic” broader class.
Table 2.
Micro-averaged performance scores for all classification tools, over the three broader classes in the VCEP-HL data set.
Best values of a performance score, across all classification tools, are shown in bold.
Fig 9.
Activation frequency ratios for VCEP-HL data set.
Log ratios calculated for each of the three classes classified by the VCEP-HL: (A) “Benign” broader class, (B) “VUS” class and (C) “Pathogenic” broader class.
Fig 10.
ROC curves and AUC scores of all classification tools for MHH data set.
(A) Prediction of “Benign” broader class versus “Pathogenic” broader class and “VUS” class (B) Prediction of “Pathogenic” broader class versus “Benign” broader class and “VUS” class (C) Prediction of “VUS” class versus “Benign” broader class and “Pathogenic” broader class.
Fig 11.
Precision-recall curves and AUC scores of all classification tools for the MHH data set.
(A) Prediction of “Benign” broader class versus “Pathogenic” broader class and “VUS” class (B) Prediction of “Pathogenic” broader class versus “Benign” broader class and “VUS” class (C) Prediction of “VUS” class versus “Benign” broader class and “Pathogenic” broader class.
Table 3.
Micro-averaged performance scores for all classification tools, over the three broader classes in the MHH data set.
Best values of a performance score, across all classification tools, are shown in bold.
Fig 12.
Activation frequency ratios for MHH data set.
Log ratios calculated for each of the three classes classified by the MHH manual curators: (A) “Benign” broader class, (B) “VUS” class and (C) “Pathogenic” broader class.