Feasibility of genomic prediction for brown rot (<i>Monilinia</i> spp.) resistance in peach

Wanfang Fu; Cassia da Silva Linge; John Mark Lawton; Ksenija Gasic; Wanfang Fu; Cassia da Silva Linge; John Mark Lawton; Ksenija Gasic

doi:10.48130/FruRes-2022-0002

Figures (6) Tables (2)

Figure 1. Disease severity index (DSI) distribution in the three datasets (Bolinha, Contender, Combined).
Figure 2. Number of SNPs included in genome wide association study and genomic prediction analysis for the three datasets (Bolinha, Contender and Combined). (a) Number of SNPs per chromosome. (b) Venn diagram for SNPs shared by all three datasets.
Figure 3. (a) Heat map of kinship matrix of the testing panel and, (b) population structure of the three datasets (Bolinha, Contender and Combined).
Figure 4. Predictive accuracy for five-fold cross validation of several genomic prediction models and marker-assisted selection (MAS) model in three datasets (Bolinha, Contender and Combined) for wounded (W) and non-wounded (NW) disease severity index (DSI).
Figure 5. Predictive accuracy for leave-one-family-out cross-validation of several genomic prediction models in the Bolinha and Contender datasets for wounded (W) and non-wounded (NW) disease severity index (DSI).
Figure 6. Comparison of mean genomic estimated breeding value (GEBV) between individuals with low and high field observed disease incidence (FDI).

Trait	Training set	Validation set	Model	Predictive accuracy
NW DSI	Bolinha	Contender	Bayes A	0.154
			Bayes B	0.166
			Bayes C	0.147
			Bayesian LASSO	0.156
			Bayesian Ridge	0.156
			GBLUP	0.158
			RKHS	0.128
			rrBLUP	0.154
		Other	Bayes A	0.279
			Bayes B	0.292
			Bayes C	0.307
			Bayesian LASSO	0.279
			Bayesian Ridge	0.289
			GBLUP	0.293
			RKHS	0.301
			rrBLUP	0.293
	Contender	Bolinha	Bayes A	0.06
			Bayes B	0.057
			Bayes C	0.057
			Bayesian LASSO	0.079
			Bayesian Ridge	0.066
			GBLUP	0.063
			RKHS	0.052
			rrBLUP	0.06
		Other	Bayes A	−0.129
			Bayes B	−0.139
			Bayes C	−0.107
			Bayesian LASSO	0.012
			Bayesian Ridge	−0.117
			GBLUP	−0.095
			RKHS	−0.132
			rrBLUP	−0.095
W DSI	Bolinha	Contender	Bayes A	0.157
			Bayes B	0.131
			Bayes C	0.125
			Bayesian LASSO	0.097
			Bayesian Ridge	0.121
			GBLUP	0.147
			RKHS	0.147
			rrBLUP	0.13
		Other	Bayes A	0.121
			Bayes B	0.104
			Bayes C	0.105
			Bayesian LASSO	0.11
			Bayesian Ridge	0.097
			GBLUP	0.106
			RKHS	0.116
			rrBLUP	0.106
	Contender	Bolinha	Bayes A	−0.133
			Bayes B	−0.15
			Bayes C	−0.133
			Bayesian LASSO	−0.132
			Bayesian Ridge	−0.147
			GBLUP	−0.096
			RKHS	−0.137
			rrBLUP	−0.103
		Other	Bayes A	−0.169
			Bayes B	−0.139
			Bayes C	−0.177
			Bayesian LASSO	−0.192
			Bayesian Ridge	−0.139
			GBLUP	−0.219
			RKHS	−0.142
			rrBLUP	−0.218

Table 1. Predictive accuracy of genomic prediction models for wounded (W) and non-wounded (NW) disease severity (DSI) traits when using the Bolinha or Contender datasets as the training population.

	BLUP	rrBLUP	GBLUP	BayesA	BayesB	BayesC	BayesLasso	BayesRidge	RKHS
FDI_BLUP	−
rrBLUP	0.550**	−
GBLUP	0.549**	1.000**	−
BayesA	0.548**	0.995**	0.995**	−
BayesB	0.549**	0.998**	0.999**	0.991**	−
BayesC	0.550**	0.998**	0.998**	0.998**	0.995**	−
BayesLasso	0.537**	0.994**	0.994**	0.984**	0.997**	0.989**	−
BayesRidge	0.553**	0.995**	0.994**	0.999**	0.990**	0.998**	0.982**	−
RKHS	0.552**	0.991**	0.991**	0.994**	0.986**	0.994**	0.978**	0.994**	−
** − Correlation detected at the significant level of 0.01.

Table 2. Spearman's correlation between BLUP of field observed flesh disease incidence (W_FDI) and genomic estimated breeding value obtained from genomic prediction models.