Rule-based induction method for haplotype comparison and identification of candidate disease loci

Table 3 Summary of the results from the simulated dataset using different thresholds

		Window sizes
	Threshold	20	30	50	100	120	150	180	200
Percentage of results that included the	3	36	80	96	100	100	100	100	100
mutation (%)	4	23	68	91	100	100	100	100	100
	5	20	62	88	100	100	100	100	100
	6	19	62	88	100	100	100	100	100
	7	19	62	88	100	100	100	100	100
	8	19	62	88	100	100	100	100	100
	9	19	62	88	100	100	100	100	100
	10	19	62	88	100	100	100	100	99
	11	18	62	87	99	99	99	98	99
Percentage of mutation loci in the top hit	3	8	3	18	29	30	37	39	42
(%)	4	13	3	19	28	30	35	37	38
	5	15	3	19	29	31	34	38	41
	6	21	3	19	31	35	39	43	45
	7	21	3	23	33	39	43	48	51
	8	21	5	27	37	43	45	50	54
	9	26	11	31	45	49	54	60	64
	10	26	13	39	55	58	65	68	70
	11	33	26	52	74	80	85	87	87
Mean length of haplotypes (number of	3	18	49	166	470	603	788	849	912
SNPs)	4	18	39	138	375	485	581	634	676
	5	18	37	129	350	416	479	532	572
	6	19	37	129	344	408	467	520	559
	7	19	37	128	344	408	467	520	559
	8	19	37	128	345	411	470	524	561
	9	19	37	128	344	405	458	508	544
	10	19	37	128	344	403	455	500	538
	11	20	37	125	334	383	430	468	491

The threshold for the number of cases was varied in all window sizes. The results where six cases were required to share the haplotype (rows in bold), which corresponds to the lymphoma study, were compared with the BEAGLE results. The percentages of datasets that included the mutation and the percentage of mutated haplotypes being the top hit were calculated from the 100 simulated datasets. The mean length of haplotypes was also calculated from the 100 simulated datasets.

ISSN: 1756-994X