Additional file 1 of X-CNV: genome-wide prediction of the pathogenicity of copy number variations

Additional file 1. Table S1. The sample information of the natural population collected from dbVar and DGV. Table S2. The 19,575 CNVs from dbVar used for model training. Table S3. The 8966 CNVs used for model validation and 22,976 likely pathogenic, likely benign, or uncertain CNVs used to quantitively measure the pathogenic effect from ClinGen and DECIPHER. Table S4. The 30 predictive features used by XGBoost algorithm. Table S5. The 1666 CNVs with a definite phenotype from DECIPHER database. Table S6. 32 CNV losses with a definite phenotype from DECIPHER, which were located in 8 tumor-suppressing cancer predisposition genes (CPGs), including APC, CDKN2A, CHEK2, NF1, NF2, RB1, TP53, and TSC2. Table S7. The coverage of the unified CNVs on the chromosomes of the human genome.