A High-Throughput Computational Framework for Identifying Significant Copy Number Aberrations from Array Comparative Genomic Hybridisation Data

Table 1

Comparison of regions of interest identified by swatCGH following four methods of DNA segmentation. The values shown are derived from 3-probe window analysis of the GBM dataset, using adaptive thresholding to limit CNAs to 10% of the genome. For each segmentation method data is provided for regions of copy number gain, regions of copy number loss, and for the total CNA. Italic columns represent findings for CRIs using unfiltered data, while roman columns represent data for MRIs filtered for significance using amplitude-based prioritization (𝑃<0.1).

	All regions		Filtered regions (𝑃<0.1)
	Number CRIs	Proportion CNA	Number CRIs	Proportion CNA
	BioHMM
Gain	253	7.34%	69	1.33%
Loss	227	3.62%	44	0.56%
Total	480	10.96%	113	1.89%
Gain : Loss		2.03		2.40
	GLAD
Gain	66	18.91%	11	1.04%
Loss	65	12.17%	7	0.18%
Total	131	31.08%	18	1.22%
Gain : Loss		1.55		5.67
	DNAcopy
Gain	99	9.81%	13	0.81%
Loss	118	15.36%	19	0.99%
Total	217	25.17%	32	1.80%
Gain : Loss		0.64		0.82
	HomHMM
Gain	275	5.48%	46	0.93%
Loss	216	3.39%	24	0.29%
Total	491	8.87%	70	1.23%
Gain : Loss		1.62		3.20