An efficient unified model for genome-wide association studies and genomic selection

Table 4 Comparison of the mapping precision between stepwise linear mixed model (StepLMM) and other methods based on QTLMAS16 data with 50 simulated QTL

Method	Number of false positives				Number of true QTL				Ratio^a
Method	Trait1	Trait2	Trait3	Total	Trait1	Trait2	Trait3	Total	Ratio^a
RR_YD^b	9	15	5	29	8	6	8	22	0.43
GRAMMAR^c	0	0	0	0	2	3	5	10	1.00
RHM20^d	1	0	0	1	6	4	7	17	0.94
RF_YD^e	3	2	0	5	3	3	5	11	0.69
LDLA^f	3	3	1	7	6	2	5	13	0.65
LA^g	4	3	1	8	0	1	2	3	0.27
StepLMM	0	0	0	0	5	4	2	11	1.00

^aCalculated as the ratio of the number of detected true QTL to the number of all detected QTL
^bRidge regression on actual yield deviations [29]
^cGenome-wide rapid association using mixed model and regression [30]
^dRegional heritability mapping (20 SNPs) [32]
^eRandom forest with yield deviations [39]
^fLinkage disequilibrium and linkage analysis [31]
^gLinkage analysis [33]

ISSN: 1297-9686