Table 1

Comparison between optimal objective and hierarchical objective value.

Data Set

Dimension

Optimal Objf

HC Objf

HC (Opt. Order) Objf

Metabolite

Rows

2,662.8

3,783.2 (29.6%)

3,550.2 (25.0%)

Concentration [39]

Columns

1,753.0

2,044.3 (14.2%)

1,865.8 (6.0%)

Colon

Rows

26,602.6

40,878.5 (34.9%)

35,637.0 (25.4%)

Cancer [44]

Columns

32,174.0

43,627.2 (26.3%)

39,138.3 (17.8%)

Breast

Rows

27,613.8

38,572.5 (28.4%)

36,182.4 (23.7%)

Cancer [48]

Columns

42,711.4

49,064.7 (12.9%)

48,553.1 (12.0%)

Yeast

Rows

82,162.4

120,429.0 (31.8%)

111,612.1 (26.4%)

Segregant [50]

Columns

124,441.0

154,353.3 (19.4%)

154,353.3(19.4%)

"Objf" denotes the objective function for the squared difference metric (see Eq. 3) and hierarchical clustering (using the Euclidean metric), with and without optimal leaf ordering. "HC" corresponds to hierarchical clustering [4] and "HC (Opt. Order)" corresponds to the optimal leaf ordering of hierarchical clustering using TreeArrange [41].

DiMaggio et al. BMC Bioinformatics 2008 9:458   doi:10.1186/1471-2105-9-458