Quantitative utilization of prior biological knowledge in the Bayesian network modeling of gene expression data

BMC Bioinformatics

Table 3 The improvement in network modeling with the addition of prior knowledge

Data set	Simulated data	Yeast cell cycle study, benchmark from BIND	Yeast cell cycle study, benchmark from ChIP-chip	Mouse pancreas study
Number of genes	76	107	107	36
Number of established regulations	124	114	190	24
Number of possible regulations	76*75 = 5700	107106/2 = 5671	9*106 = 954	36*35 = 1260
Number of known regulations recovered with (without) prior knowledge	21 (14)	26 (13)	23 (11)	12 (6)
Total number of regulations predicted, with (without) prior knowledge	503 (440)	436 (387)	58 (33)	322 (297)
Improvement over plain BN	χ ² = 0.36, p~0.54	χ ² = 2.28, p < 0.13	χ ² = 0.04, p~0.84	χ ² = 0.98, p~0.32
Improvement: over random selection	χ ² = 7.32, p < 0.01	χ ² = 24.5, p < 0.001	χ ²= 6.71, p < 0.01	χ ² = 2.87, p < 0.09
Plain BN over random selection	χ ² = 1.58, p~0.2	χ ² = 2.42, p~0.11	χ ² = 1.6, p~0.2	χ ² = 0.01, p~0.8

* We ignored edge direction with comparing to BIND since it contains both directed and undirected interactions.

ISSN: 1471-2105