Fig. 2: Heritability estimates and prediction accuracy of the building blocks of dynamicGP on the maize dataset. | Nature Plants

Fig. 2: Heritability estimates and prediction accuracy of the building blocks of dynamicGP on the maize dataset.

From: Predicting plant trait dynamics from genetic markers

Fig. 2: Heritability estimates and prediction accuracy of the building blocks of dynamicGP on the maize dataset.

a, The heritability estimates for each entry of the matrices from algorithm 2 used as building blocks of dynamicGP, as obtained by genomic-relatedness-based restricted maximum likelihood (GREML). Here, \(\widetilde{{{A}}}\) denotes the matrix resulting from the rank-reduced proper orthogonal decomposition-projected representation in algorithm 2; U, Σ and V are obtained from the singular value decomposition of X1; R and Q are matrices obtained from the Schur decomposition of \(\widetilde{{{A}}}\), while Φ collects the projected modes from DMD of the operator A. b, The prediction accuracy of the building blocks of dynamicGP using RR-BLUP in 20 iterations of a 5-fold cross-validation.

Source data

Back to article page