A two-step multiple-marker strategy for genome-wide association studies

BMC Proceedings

Table 2 Comparison of multiple marker and single marker strategies

	Correction for multiple testing	Sensitivity^a	Specificity^a	FDR^a	% of all TL detected^b
		FBAT-LC
Multiple markers^c	None	0.97 (0.08)	0.95 (0.03)	0.37 (0.17)	0.56 (0.02)
	Benjamini & Hochberg	0.95 (0.10)	0.99 (0.09)	0.03 (0.09)	0.55 (0.02)
	Bonferroni	0.94 (0.11)	0.99 (0.09)	0.03 (0.09)	0.55 (0.03)
		Single-marker FBAT
Single marker^d	None	0.97 (0.09)	0.78 (0.03)	0.83 (0.02)	0.56 (0.02)
	Benjamini & Hochberg	0.95 (0.11)	0.87 (0.01)	0.74 (0.02)	0.55 (0.02)
	Bonferroni	0.95 (0.11)	0.87 (0.01)	0.74 (0.03)	0.55 (0.02)

^aMean (standard deviation) of sensitivity, specificity, and false-discovery rate of FBAT-LC (multiple markers) and single-marker FBAT computed over 50 replicates of family data.
^bMean (standard deviation) of the proportion of all trait loci (TL) selected in Step 1 and detected in Step 2 by FBAT at the 5% level over 50 replicates.
^c50 regions were selected in Step 1 using the local score method.
^d115 SNPs were selected in Step 1 (see text) using the Pearson chi-square applied to case-control genotypic contingency table.

ISSN: 1753-6561