Table 1.

Comparison of performance measures. For evaluation of our method, we compared our predicted TF-gene functional pairs (fourth column in bold, 'Our_final') with ChIP-chip results with annotated genes only (first column, 'ChIP-chip'), two prescription steps of our method (second and third columns, 'CM' and 'FI_TM'), and two other previous algorithms (fifth and sixth columns, 'GRAM' and 'MA-Networker'). Two performance measures were calculated, PPV and SNST for two reference datasets (see Methods). CM = all TF-gene pairs from coherent modules; FI_TM = TF-gene pairs from functional intersection among the initial putative transcriptional modules from ChIP-chip; lit = literature reference; con_mot = conserved motif reference; N_pairs = number of TF-gene pairs; N_genes = number of genes in the pairs; N_TFs = number of TFs in the pairs. Further comparison analysis is performed in Additional file 1. See the main text for details.

	ChIP-chip	CM	FI_TM	Our_final	GRAM	MA-Networker
PPV (%)(lit; con_mot)	4.6; 32.7	6.0; 35.8	18.2; 24.5	13.6; 48	6.3; 24.6	6.5; 38.6
SNST (%)(lit; con_mot)	13.7; 40.2	4.2; 10.5	1.7; 0.9	2.0; 2.9	7.9; 12.8	6.9; 16.8
N_pairs	3598	857	110	177	1518	1272
N_genes	1837	393	44	66	655	989
N_TFs	95	24	30	18	69	36