Fig. 1
From: X-CNV: genome-wide prediction of the pathogenicity of copy number variations
Workflow of X-CNV model training and validation. The model was trained based on the XGBoost algorithm using 30 predictive features of 5315 pathogenic and 14,260 benign CNVs from dbVar and was validated in 4893 pathogenic and 4073 benign CNVs from ClinGen and DECIPHER. The features were categorized into four types, including universal annotation, genome-wide annotation, coding annotation, and non-coding annotation. The allele frequency (AF) of CNVs was calculated based on the unified CNVs from DGV and dbVar