Data- and expert-driven rule induction and filtering framework for functional interpretation and description of gene sets

Table 1 Comparison of different logical rule generation methods and different parameter settings

	S01	S02	S03	S04(1)	S04(2)	S04(3)	S04(4)	S04(5)	S04(6)
No. of rules before filtering	3812	3812	3812	110	110	110	110	110	110
No. of output rules	3812	32	32	9	10	7	19	110	14
No. of rules with expert terms	1465	15	11	9	10	7	19	110	14
Coverage	82	82	82	64	64	64	64	64	64
Avg. p-value	0.018	0.017	0.014	0.009	0.012	0.013	0.019	0.016	0.014
Avg. precision	0.74	0.78	0.77	0.81	0.78	0.7	0.68	0.71	0.72
Avg. coverage	0.14	0.15	0.15	0.16	0.16	0.16	0.16	0.15	0.17
Avg. GO Level	4.06	4.18	3.7	4.95	4.84	5.8	4.66	4.51	4.7
Positive coverage	18	18	18	14	14	14	14	14	14
Negative coverage	57	35	36	11	12	13	19	20	14
Positive coverage - expert rules	14	13	11	14	14	14	14	14	14
Negative coverage - expert rules	28	10	11	11	12	13	19	20	14
Avg. no. of descriptors	3.57	3.19	3.53	2.33	2.5	1.43	2.47	2.66	2.36
Avg. no. of expert term per rule	0.41	0.47	0.38	1.44	1.4	1.14	1.53	1.35	1.29
Number of distinctive expert terms	19	8	6	9	9	8	13	19	11

S01 – RuleGO method without filtering procedure, S02 – standard RuleGO method with applied filtering, S03 – filtering using UTA approach, S04 – new rule generation approach using seed terms. Description of different Q Compound measure and filtering setting for S04(1)-S04(6) is presented in Table 2

ISSN: 2041-1480