BMC Plant Biology

Table 3 Prediction accuracy of six genomic selection (GS) models for five fruit traits based on the confident 31,142 SNPs in three training populations

From: Prediction accuracy of genomic estimated breeding values for fruit traits in cultivated tomato (Solanum lycopersicum L.)

GS model^a		Training population	Prediction accuracy^b
GS model^a		Training population	FW^c	FWt	FH	PT	Brix
Parametric	RR-BLUP	TGC1	0.822	0.867	0.821	0.856	0.670
		TGC2	0.748	0.766	0.687	0.618	0.776
		Combined	0.758	0.802	0.719	0.765	0.736
	BA	TGC1	0.824	0.868	0.821	0.856	0.673
		TGC2	0.744	0.761	0.686	0.624	0.772
		Combined	0.775	0.804	0.715	0.765	0.734
	BL	TGC1	0.816	0.861	0.815	0.853	0.686
		TGC2	0.734	0.779	0.679	0.623	0.779
		Combined	0.766	0.799	0.708	0.748	0.739
Non-parametric	RKHS	TGC1	0.828	0.870	0.822	0.859	0.682
		TGC2	0.758	0.775	0.700	0.625	0.784
		Combined	0.777	0.813	0.738	0.773	0.746
	SVM	TGC1	0.804	0.851	0.797	0.860	0.690
		TGC2	0.755	0.772	0.669	0.594	0.797
		Combined	0.778	0.808	0.723	0.767	0.765
	RF	TGC1	0.835	0.865	0.810	0.866	0.702
		TGC2	0.780	0.791	0.641	0.643	0.778
		Combined	0.812	0.834	0.728	0.807	0.751

^aRidge regression-best linear unbiased prediction (RR-BLUP), BayesA (BA), Bayesian LASSO (BL), reproducing kernel Hilbert space (RKHS), support vector machine (SVM), random forest (RF)
^bPrediction accuracy was estimated using the Pearson correlation coefficients between genomic estimated breeding values (GEBVs) and observed phenotypes
^cFW (fruit weight), FWt (fruit width), FH (fruit height), and PT (pericarp thickness)

Back to article page

ISSN: 1471-2229

Contact us

General enquiries: journalsubmissions@springernature.com