Skip to main content

Advertisement

Table 4 Sensitivity of each detection program when other experimental result is regarded as golden standard

From: Comparative analysis of copy number variation detection methods and database construction

Programs Korbel[6](PE sequencing) Kidd[7](PE sequencing) Kidd[7]-validation Tuzun[8](PE sequencing) Conrad[25]**(Mendelian consistency) Shaikh[26]** (Illumine Data) Redon[11](Tiling array) Perry[27](array CGH) Conrad[28](Tiling array) Park[29](array-CGH and massively parallel sequencing)
PennCNV 2.2*
(2.1)
0.5
(0.7)
1.2
(1.5)
5.2
(7.8)
23.7 37.0
[37.1]
1.2
(1.1)
18.1
(22.3)
26.2
(24.5)
27.0
(33.0)
Birdseye
(part of Birdsuite)
2.4
(2.4)
0.6
(1.0)
1.4
(2.1)
7.7
(13.9)
30.0 40.7
[44.4]
0.8
(0.9)
13.4
(14.0)
23.5
(24.1)
27.9
(36.9)
Birdsuite 5.5
(6.1)
2.0
(4.3)
5.5
(8.8)
17.2
(24.8)
46.1 51.9
[70.4]
2.2
(3.2)
25.6
(30.0)
35.4
(35.9)
42.9
(56.2)
DNAcopy 1.7
(1.6)
0.4
(0.9)
1.1
(1.9)
5.2
(9.0)
20.4 44.4
[48.1]
0.7
(1.2)
9.9
(11.8)
15.3
(17.9)
19.1
(28.7)
CGHseg 0.5
(0.3)
0.1
(0.3)
0.3
(0.6)
1.5
(2.7)
18.2 29.6
[29.6]
0.4
(0.7)
7.0
(6.4)
7.7
(9.3)
9.6
(12.6)
  1. * Sensitivity of all data, () represents deletion-only data.
  2. Sensitivity: Common data between "other experimental results" and "each CNV detection program's results" divided by "other experimental results".
  3. Commonly utilized individuals between this study and other experiment are used for calculating sensitivity and specificity.
  4. CNV segments with overlap > 80% are regarded as commonly detected segments.
  5. Kidd-validation: Kidd results that are also detected using other experimental methods in the paper7.
  6. **Conrad and Shaikh data include only deletion data.
  7. [] represents ratios with overlapping ratio > 30% and CNV length > 10 kb.