FastKnock: an efficient next-generation approach to identify all knockout strategies for strain optimization

Table 1 The number of linear programming problems (LPs) solved by the FastKnock algorithm compared to an exhaustive search of the preprocessed search space (Strain0 in CM2 cultivation medium)

			Single	Double	Triple	Quadruple	Quintuple
Strain0 in CM2	iJR904	Exhaustive search	208	21,528	1,478,256	75,760,620	3,091,033,296
		FastKnock	41	820	11,613	125,815	1,178,030
		% Reduction	80.29	96.20	99.22	99.84	99.97
	iAF1260	Exhaustive search	315	49,455	5,159,805	402,464,790	25,033,309,938
		FastKnock	57	1,506	25,985	348,966	4,058,061
		% Reduction	81.91	96.96	99.50	99.92	99.99
	iJO1366	Exhaustive search	385	73,920	9,437,120	901,244,960	68,674,865,952
		FastKnock	58	2,038	43,565	732,315	10,822,208
		% Reduction	84.93	97.24	99.53	99.91	99.98
	iML1515	Exhaustive search	403	81,003	10,827,401	1,082,740,100	86,402,659,980
		FastKnock	61	2193	58,750	1,674,010	25,489,714
		% Reduction	84.87	97.30	99.46	99.85	99.98

ISSN: 1475-2859