Evaluation of reaction gap-filling accuracy by randomization

BMC Bioinformatics

Table 3 GenDev Technique C: Results for 2400 computational experiments when using the General Development Mode (GenDev) with a dynamic biomass reaction constrained to a minimum flux of 10⁻³ but not as part of the objective function

(1) # rxns removed	1		2		5		10		20		50
(2) Rxns suggested are identical to rxns removed	63	na	28	na	6	na	0	na	0	na	0	na
	59	54	24	27	5	7	0	1	0	0	0	0
(3) No rxns suggested	25	na	12	na	0	na	0	na	0	na	0	na
	25	25	12	12	0	0	0	0	0	0	0	0
(4) Rxns suggested are a strict subset of rxns removed	0	na	34	na	38	na	23	na	7	na	0	na
	0	0	34	33	35	49	19	37	8	20	0	0
(5) # rxns suggested is equal to # rxns re-moved	7	na	11	na	15	na	6	na	5	na	2	na
	9	9	14	13	19	10	10	3	3	0	0	0
(6) # rxns suggested is less than # rxns re-moved	0	na	7	na	25	na	59	na	82	na	97	na
	0	0	6	9	29	32	64	59	88	80	99	100
(7) # rxns suggested is greater than # rxns removed	5	na	8	na	16	na	12	na	6	na	1	na
	7	12	10	6	12	2	7	0	1	0	1	0
(8) # cases where no solution was found	0	na	0	na	0	na	0	na	0	na	0	na
	0	0	0	0	0	0	0	0	0	0	0	0
Precision (%)	87	na	80	na	76	na	78	na	77	na	74	na
	83	75	77	76	76	82	78	87	78	87	75	88
Average precision (%)						79	na
						78	83
Recall (%)	67	na	56	na	61	na	59	na	59	na	59	na
	53	59	53	51	60	60	59	60	58	60	58	60
Average recall (%)						60	na
						57	58
# Solutions that do not show growth						0	na
						0	1

ISSN: 1471-2105