Using local gene expression similarities to discover regulatory binding site modules

BMC Bioinformatics

Table 1 Evaluation with Gene Ontology

Study	Molec. function	Biol. Process	Cell. comp.
Previously published studies
Segal et al. [26], Cell Cycle, 17 modules	0.353	0.353	0.471
Segal et al. [25], 48 modules	0.250	0.417	0.312
Beer et al. [24] 49 modules	0.426	0.617	0.583
Hvidsten et al. [4] Cell cycle 109 rules	0.308	0.462	0.410
Present method
corr. 0.6, 1316 rules	0.335 (0.349)	0.479 (0.510)	0.422 (0.406)
corr. 0.7, 807 rules	0.522 (0.395)	0.657 (0.532)	0.621 (0.551)
corr. 0.75, 643 rules	0.453 (0.443)	0.526 (0.471)	0.711 (0.480)
corr. 0.8, 487 rules	0.427 (0.361)	0.491 (0.304)	0.719 (0.476)

Comparison of the Gene Ontology-based scores with other studies. For each method, the presented number represents the fraction of significant gene groups among all gene groups provided by that method (significance is defined as Bonferroni corrected p-value being below 0.01). For the present method, the results that score higher than all other studies are shown in bold. Correlation thresholds used in the clustering procedure are shown in the left column. Numbers in parentheses denote scores obtained for all genes (i.e. both genes with and without expression profiles).

ISSN: 1471-2105