Table 3

Comparison of various algorithms on the 188× coverage breast cancer genome

Sample PD4120a


Algorithm

% normal admixture

Clonal (% tumor purity)a

Subclonal (%)a


THetA, n = 2

34.3%

Del: 1p, 4q, 13, 16q, 22q

-

(segmentation)

+1: 1q

(65.7%)


THetA, n = 2 (chromosome arms)

38.3%

Del: 1p, 4q, 13, 16q, 22q

-

+1: 1q

(61.7%)


CNAnormb

(chromosome arms)

32.8%

Del: 1p, 4q, 13, 16q, 22q

-

+1: 1q

(67.2%)


ASCATc

34%

Del: 1p, 4q, 13, 16q, 22q

-

(virtual SNP array)

+1: 1q, 17q, 18, 19, 20

(66.0%)


ABSOLUTEd (segmentation)

35%

(65.0%)

-


THetA

28.0%

Del: 1p, 4q, 16q, 22q12.2-

Del: 13, 22q11.2-12.1

n = 3

13.3

+1: 1q

(72.0%)

(61.9%)

Del: 8, 11, 12, 14, 15

(10.1%)

Del: 2, 7, 4p, 6,

9, 18, 21


Nik-Zainal et al. (2012)

30%

Del: 4q

Del: 13, t(1;22)

[25]

+1: 1q

(47.6%)

(70.0%)

Tetraploid with:

Del(-2): 2, 7

Del(-1): 6, 8, 9, 11, 12,

14, 15, 18, 21

(9.8%)


a Entries in bold are differences between THetA and [25]. Entries in italics were not input to the n = 3 THetA analysis but were inferred using THetA's output.

b When CNAnorm was run using BIC-Seq intervals the normal admixture was estimated at 6.7%, therefore we report results from CNAnorm using chromosome arms. CNAnorm does not return integer copy numbers - and thus we report aberrations where the returned copy number was within 0.15 of the nearest integer, other aberrations were considered inconclusive.

c For ASCAT we use virtual SNP array data as input. ASCAT performs its own segmentation; we list only the large aberrations.

d We report here the maximum likelihood solution returned by ABSOLUTE when considering only karyotypes. When considering only somatic copy number aberrations or a combination, ABSOLUTE infers a tetraploid solution. For this sample, ABSOLUTE returns copy numbers for only a subset of the input intervals, so we do not report specific copy number aberrations predicted.

Oesper et al. Genome Biology 2013 14:R80   doi:10.1186/gb-2013-14-7-r80

Open Data