Table 3.

Performance results of all tested color-space read aligners on simulated data with error threshold of two mismatches

Dataset	Mapper	Time (min)	Map locations	Reads mapped (%)
Set 1	drFAST-DP	65	138 715 908	100
	drFAST-CT	40	137 483 484	100
	BFAST	88	8 803 840	96.1
	Bowtie v = 2	17	25 581 176	99.4
	Bowtie n = 2	67	168 307 651	99.4
	SHRiMP	414	13 961 155	99.8
	SOCS	45	13 357 519	100
	Mapreads	50	55 569 848	100
	PerM	17	14 441 796	96.2
Set 2	drFAST-DP	42	37 652 313	100
	drFAST-CT	26	36 458 468	100
	BFAST	101	8 098 581	98.0
	Bowtie v = 2	13	9 738 234	60.8
	Bowtie n = 2	31	57 550 920	61.9
	SHRiMP	519	11 977 512	99.8
	SOCS	90	12 909 860	100
	Mapreads	31	21 749 155	100
	PerM	15	12 679 070	98
Set 3	drFAST-DP	47	76 588 622	100
	drFAST-CT	32	75 970 911	100
	BFAST	105	8 982 132	97.4
	Bowtie v = 2	16	11 030 554	49.4
	Bowtie n = 2	43	70 508 835	51.66
	SHRiMP	472	11 859 215	99.8
	SOCS	96	9 780 960	100
	Mapreads	37	29 799 473	100
	PerM	15	13 140 561	97.5

In the case of PerM, we allowed for mapping with up to four mismatches as recommended by its developers, yet its sensitivity failed to reach 100%. Reads are simulated from human reference genome build 35 (chromosome 1). Set 1: no errors; Set 2: color errors; Set 3: substitutions. Values in bold show alignments with 100% sensitivity, higher value implies more sensitivity in the reported alignment.