Fig 1.
White mold phenotypic data distribution transformed using logistic regression of 465 soybean accessions tested in field and greenhouse specialized tests.
Fig 2.
Relationship between predictive ability and the number of SNP markers using a training population of 352 genotypes among 465 diverse soybean accessions tested for white mold in 2014 and 2015 field and greenhouse experiments.
Fig 3.
Predictive ability for white mold reaction phenotyped for WM in field and greenhouse screening in 2014 and 2015 using RR-BLUP for differing training population sizes using 5 k SNP markers using 465 diverse soybean accessions.
Table 1.
Predictive ability and standard deviation reported using RR-BLUP from white mold evaluation experiments in field and greenhouse (2014 and 2015), when one environment was used to train the model and validated on another environment.
The predictive ability (estimated by 10 fold cross-validation) using the same experiment for training and validation population was used as control.
Fig 4.
Scheme demonstrating the use of genomic selection models in training, validating, and testing sets.
Top– 5% and 10% most resistant accessions for WM found in the present study (F2015); Bottom– 5% and 10% most susceptible accessions for WM found in the present study (F2015); USDA_Top– 5% most resistant accessions found in USDA soybean germplasm collection; Test–Resistant accessions previously reported by other researches. GWS = Genome wide selection, GEBV = Genomic estimated breeding value, WM = white mold.