%0 Generic %A Stewart-Brown, Benjamin %A Song, Qijian %A Vaughn, Justin %A Li, Zenglu %D 2019 %T Supplemental Material for Stewart-Brown et al., 2019 %U https://gsajournals.figshare.com/articles/dataset/Supplemental_Material_for_Stewart-Brown_et_al_2019/8121251 %R 10.25387/g3.8121251.v1 %2 https://gsajournals.figshare.com/ndownloader/files/15139673 %2 https://gsajournals.figshare.com/ndownloader/files/15139661 %2 https://gsajournals.figshare.com/ndownloader/files/15139664 %2 https://gsajournals.figshare.com/ndownloader/files/15139670 %2 https://gsajournals.figshare.com/ndownloader/files/15139667 %2 https://gsajournals.figshare.com/ndownloader/files/15139676 %2 https://gsajournals.figshare.com/ndownloader/files/15139649 %2 https://gsajournals.figshare.com/ndownloader/files/15139655 %2 https://gsajournals.figshare.com/ndownloader/files/15139658 %K Genomic Selection %K RR-BLUP %K Seed Composition %K Seed Yield %K Soybean %K Breeding %K Protein %K Oil %K Prediction %K Crop and Pasture Improvement (Selection and Breeding) %X Table S1 displays the effect of training set size on prediction ability when performing cross-validation across the entire genomic selection dataset. Table S2 displays the effect of marker density on prediction ability when performing cross-validation across the entire genomic selection dataset. Table S3 displays the effect of training set size on prediction ability when performing cross-validation to predict individual bi-parental families using the within population method versus across population method (Prediction ability averaged across Pop1-4). Table S4 displays the effect of training set size on prediction ability when performing cross-validation to predict individual bi-parental families using the within population method versus the across population method (Prediction ability displayed for each individual validation population). Figure S1 contains kernel density plots of BLUP values for phenotypic traits of interest. Figure S2 contains boxplots of observed BLUP values for traits of interest, broken out by populations and pedigrees used for genomic prediction. Figure S3 displays the effects of training set size on predictive ability when contrasting the within population method vs. the across population method. All data and code required to replicate the analyses are available in Files S1-5. File S1 contains the raw phenotypic data for calculation of BLUP values. File S2 contains the phenotypic BLUP values used for each method of GS. Worksheet 1 provides additional information for each genotype while Worksheets 2 and 3 contain information used for GS. Supplemental Data File S3 and S4 contains the genotypic data files used for prediction of seed yield and protein/oil content for each method of GS. Worksheet 1 of each file provides additional information for each genotype while Worksheets 1-10 contain genotypic data for the following marker densities: all SNPs, tag SNPs, half tag SNPs, 4th tag SNPs, and 8th tag SNPs. Odd sheets contain extra information while even sheets contain data used for GS. File S5 provides the r code used for calculation of predictive abilities which can be adapted to test all methods. %I GSA Journals