Table 2. Comparison of reference genomic sequence datasets for mapping captured reads.

			Numbers of SNPs by allele frequency (AF)
Reference	Min. variant reads	SNP type	0.05–0.1	0.1–0.2	0.2–0.3	0.3–0.4	0.4–0.5	0.5–0.6	0.6–0.7	0.7–0.8	0.8–0.9	0.9–0.10	Total	Total (AF>0.1)
CSS	3	EMS	427	163	64	56	46	30	23	19	18	257	1103	676
		NON-EMS	294	113	28	11	1	1	0	1	0	4	453	159
CSS	8	EMS	13	22	31	38	35	26	20	18	17	246	466	453
		NON-EMS	16	3	3	1	0	0	0	0	0	4	27	11
Ensembl	3	EMS	923	419	127	74	53	27	26	16	18	253	1936	1013
		NON-EMS	660	349	91	32	3	1	0	1	0	3	1140	480
Ensembl	8	EMS	24	33	34	43	34	22	23	16	17	242	488	464
		NON-EMS	17	1	5	1	0	0	0	1	0	3	28	11
Ensembl-RM	3	EMS	426	163	61	54	47	22	27	17	16	231	1064	638
		NON-EMS	279	130	31	8	2	0	1	1	0	2	454	175
Ensembl-RM	8	EMS	17	25	32	40	37	19	24	17	16	221	448	431
		NON-EMS	12	9	2	1	0	0	0	0	0	2	26	3

Reads were mapped with Novoalign using parameter t = 60, equivalent to a mismatch setting of approximately 2. Novoalign hard clipping option was used with a base quality 15. Reads were filtered to remove those with a mapping score less than 20. References used were the full IWGSC chromosome arm survey (“CSS”), the Ensembl v21 subset of CSS (“Ensembl”) or a repeat-masked version of the latter (“Ensembl-RM”). Minimum total read coverage was 8, minimum SNP read coverage 3 or 8, and minimum SNP base quality of 20.