Analysis of multiple phenotypes in genome-wide genetic mapping studies

BMC Bioinformatics

Table 1 Top five SNPs discovered using the four methods

Five phenotypes: 3 IQ + 2 memory measurement;				$\bar{r}$ = 0.59;	n = 118
One-by-one*	rs3738401	rs9822602	Dys1578kb	rs2005976	rs821616
	0.009	0.042	0.074	0.078	0.208
Permutation	rs3738401	rs9822602	Dys1578kb	rs2005976	rs821616
	0.008	0.028	0.049	0.052	0.131
MANOVA	rs3738401	rs9822602	Dys1325kb	rs5105774	rs2604578
	0.031	0.046	0.048	0.097	0.126
PCA	rs3738401	rs5105774	rs9822602	Deletion_A7_Ms	rs720309
	0.002	0.056	0.063	0.121	0.139
Five phenotypes: 3 IQ measurement;				$\bar{r}$ = 0.72;	n = 131
One-by-one*	rs3738401	rs9822602	rs2005976	rs5751229	rs1347003
	0.037	0.04	0.084	0.103	0.149
Permutation	rs3738401	rs9822602	rs2005976	rs5751229	rs1347003
	0.029	0.031	0.061	0.068	0.101
MANOVA	Dys1325kb	rs2005976	rs9822602	rs3738401	rs5751229
	0.016	0.046	0.066	0.093	0.245
PCA	rs3738401	rs5751229	rs9822602	Deletion_A7_Ms	rs2005976
	0.015	0.095	0.099	0.106	0.122

*Bonferroni correction is applied to correct for multiple testing among phenotypes. The exact formula used is 1 - (1 - p)^m.
There are two sets of analyses, one with five moderately correlated phenotypes and the other with three highly correlated phenotypes.

ISSN: 1471-2105