Table 1.

Run time in seconds for different alignment-free approaches on six sets of real-world genomes. On the largest data set, the 14 plant genomes, ‘kmacs’ and ‘co-phylog’ did not terminate the program run. On this data set, we increased the pattern weight for ‘Multi-SpaM’ from the default value of = 10 to = 12, in order to reduce the run time. Note that ‘Multi-SpaM’, ‘FSWM’ and ‘andi’ are parallelized, so we could run them on multiple processors, while ‘kmacs’ and ‘co-phylog’ had to be run on single processors. The reported run times are ‘wall clock times’.

	FSWM		andi		co-phylog	kmacs	Multi-SpaM
	wall clock	CPU	wall clock	CPU			wall clock	CPU
27 E.coli/Shigella	710	27,075	15	291	704	5,247	603	22,185
29 E.coli/Shigella	860	32,798	16	325	533	55,736	611	21,973
25 fish mitochondria	2	8	<1	3	9	5	27	1,054
14 plants	1,107,720	28,690,489	1,808	13813	-	-	12,516	389,770
19 Wolbachia	65	2,185	3	42	113	24,961	484	15,804
8 Yersinia	91	3,333	5	34	50	1,083	183	6,182