Identification of coherent patterns in gene expression data using an efficient biclustering algorithm and parallel coordinate visualization

Table 1 Parameter settings for biclustering algorithms and post-filtering in the experiments on artificial datasets

Experiment	Algorithm/post-filtering	Parameter settings*
Artificial datasets for additive models	PA	ε = 0.5 – 2.0, N_r= 21, N_c= 5, P_o= 50
	C&C	δ = 0.04 – 0.5, α = 1.2, M = 40
	pCluster	δ = 0.5 – 1.0, N_r= 21, N_c= 5
	Post-filtering	N_r= 21, N_c= 5, P_o= 50 and M = 10
Artificial datasets for multiplicative models	PM	ε = 0.2 – 0.6, N_r= 18, N_c= 4, P_o= 25
	PAL	ε = 0.4 – 1.0, N_r= 18, N_c= 4, P_o= 25
	C&C	δ = 0.04 – 0.5, α = 1.2, M = 20
	pCluster	δ = 0.5 – 1.0, N_r= 18, N_c= 4
	Post-filtering	N_r= 18, N_c= 4, P_o= 25 and M = 5

* The definitions of parameters ε, N_r, N_cand P_ofollow those defined for the proposed algorithm, i.e. noise threshold, minimum number of rows, minimum number of columns and maximum percentage in overlap allowed in biclusters respectively. Furthermore, M denotes the maximum number of biclusters required and δ of C&C and the pCluster algorithm is defined as in the original publications [12, 25].

ISSN: 1471-2105