IRIS: a method for reverse engineering of regulatory relations in gene networks

BMC Bioinformatics

Table 4 Inference results. Column "Biological Findings" lists a short description of the features of interest and references.

Biological Findings	Observed Genes	Hidden Genes	Inference Results
Strong relationship between cyclins CLB5 and CLB6 [40] and between CLN1 and CLN2 [41]	CLB5	CLB6	P(CLB 6 = 0\|CLB 5 = 0) = 0.9851 P(CLB 6 = 1\|CLB 5 = 1) = 1.0000
	CLB6	CLB5	P(CLB 5 = 0\|CLB 6 = 0) = 1.0000 P(CLB 5 = 0\|CLB 6 = 0) = 0.8079
	CLN1	CLN2	P(CLN 2 = 0\|CLN 1 = 0) = 0.95028 P(CLN 2 = 1\|CLN 1 = 1) = 0.8449
	CLN2	CLN1	P(CLN 1 = 0\|CLN 2 = 0) = 0.6111 P(CLN 1 = 1\|CLN 2 = 1) = 0.6111
Inhibitory activity of SIC1 on cyclins CLB5 and CLB6 [42]	SIC1	CLB5	P(CLB 5 = 0\|SIC 1 = 1) = 0.7367
		CLB6	P(CLB 6 = 0\|SIC 1 = 1) = 0.7367
Inactivation of MBP1 and SWI6 causes CLB5 and CLB6 levels to fall [35]	MBP1	CLB5	P(CLB 5 = 0\|MBP 1 = 0, SWI 6 = 0) = 0.7742
	SWI6	CLB6	P(CLB 6 = 0\|MBP 1 = 0, SWI 6 = 0) = 0.8483
While CLN1 and CLN2 are active, SIC1 is degraded [43]	CLN1	SIC1	P(SIC 1 = 0\|CLN 1 = 1, CLN 2 = 1) = 0.6476
	CLN2
Inactivation activity of MBP1 and SWI6 on CLN1 and CLN2 [44]	MBP1 SWI6	CLN1	P(CLN 1 = 0\|MBP 1 = 1, SWI 6 = 1) = 0.6111
		CLN2	P(CLN 2 = 0\|MBP 2 = 1, SWI 6 = 1) = 0.6148

To infer the conditional probability reported in the "Inference Results" column, the sum-product algorithm was used, and the state of the "Observed Genes" is fixed to derive the potential behaviour of "Hidden Genes".

ISSN: 1471-2105