Table 3. Details of platform comparison.

	Arrays	No. Samples	No. Stringent CNVs	Average CNVs/Sample	Average CNV Size, kb	No. Overlapping calls	% Overlap/ Validationa	% CNVs With Insufficient Probes in Other Array	Novel CNVs/Sample
1	Agilent 1M	82	2747	33.5 ± 6.11	107	183	6	67.5	31.27 ± 5.38
	Affy 500K	82	289	3.52 ± 2.28	520	189	65	6	2.17 ± 1.22
2	Agilent 1M	262	8971	34.24 ± 6.45	97	2725	30	38	23.84 ± 5.30
	Ilmn 1M single	262	4779	18.24 ± 4.69	110	2729	57	59.12	7.82 ± 3.11
3	Agilent 1M	11	413	37.55 ± 6.76	83	99	24	39.70	28.55 ± 5.13
	Ilmn 1M duo	11	169	14.36 ± 4.32	84	108	64	67.21	5.54 ± 1.81
4	Agilent 1M	234	8270	35.49 ± 6.23	98	4439	53	5.6	16.44 ± 4.27
	Affy 6.0	234	15,315	65.45 ± 8.87	80	4701	34	77.46	45.35 ± 7.39
5	Agilent 1M	26	945	36.35 ± 6.27	108	328	35	33	23.73 ± 4.63
	Ilmn Omni 2.5M	26	1230	47.31 ± 9.77	61	380	31	81.4	32.69 ± 8.57

ASD, autism spectrum disorder; CNV, copy number variations; PDx, control cohort DNA samples.

^a% overlap/validation- refers to the percentage of CNVs that were also detected by the other array. For example, for the first platform comparison between Agilent 1M and Affy500K arrays, only 6% of the CNVs detected by Agilent 1M were also detected by Affy500K while 65% of the CNVs detected by Affy500K were also detected or validated by Agilent 1M array.