Table 2.

Description of attributes used for training separately for copy number loss and gain variants. The aggregations [“mean” and “standard deviation” (std)] are calculated for benign and pathogenic variants separately.

Attribute	CNV loss				CNV gain
	Benign		Pathogenic		Benign		Pathogenic
	Mean	Std	Mean	Std	Mean	Std	Mean	Std
Overlapped gencode elements	4.11	10.49	56.43	54.73	9.46	18.39	80.3	65.18
Protein coding genes	1.41	3.29	19.28	19.49	2.61	4.6	28.17	25.22
Pseudogenes	1.42	5.46	11.3	15.62	3.97	10.85	17.34	19.58
Micro RNA	0.07	0.45	02.08	2.98	0.23	1.18	2.91	3.38
Long non-coding RNA	0.79	1.78	16.45	16.34	1.77	3.16	21.7	18.29
Ribosomal RNA	0.0	0.2	0.02	0.15	0.0	0.05	0.03	0.2
Small nuclear RNA	0.08	0.45	1.2	2.17	0.19	0.81	1.94	2.9
Morbid genes	0.2	0.48	3.84	3.65	0.37	0.7	5.22	4.6
Disease associated genes	0.18	0.46	3.24	3.11	0.34	0.68	4.26	3.85
Haploinsufficient genes	0.02	0.14	0.48	0.69	–	–	–	–
Haploinsufficient regions	0.06	0.27	0.47	0.67	–	–	–	–
Regulatory elements	18.13	40.77	453.31	396.98	40.36	59.97	559.55	424.5
Enhancers	3.35	7.55	75.64	72.11	6.7	11.01	82.97	65.57
Open chromatin regions	03.02	6.88	58.18	53.91	6.43	10.76	66.38	51.71
Promoters	1.16	3.43	35.69	37.78	2.88	5.3	51.44	49.52
Promoter flanking regions	3.41	8.43	99.24	92.16	7.72	12.95	118.6	91.8
CTCF binding sites	5.95	15.4	153.4	146.64	13.77	23.12	197.61	168.01
TF binding sites	1.11	3.38	26.93	35.99	2.6	5.47	36.9	44.69
Manually curated regulatory elements	0.12	0.61	4.25	4.66	0.26	0.73	5.64	5.28
Triplosensitive regions	–	–	–	–	0.02	0.17	0.62	0.74