> For the complete documentation index, see [llms.txt](https://multiplemethodgs.gitbook.io/MMGS_tutorial_v1/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://multiplemethodgs.gitbook.io/MMGS_tutorial_v1/readme.md). # Description A comprehensive R package, multiple-environments multiple methods genomic selection (MMGS), developed by Mingjia Zhu, integrates the polygenic environmental interaction (PEI) and Reaction Norm (RE) methods along with 15 prediction models that include difference prediction estimated methods contains parametic, semi-parametric and non-parametric. RN model includes four steps: (1) Using CERIS algorithm (Guo, 2021) to identify an environmental index that explained the largest proportion of phenotypic variation. (2) Regressing the observed phenotypes on the identified environmental index to obtain an intercept and a slope estimate for each tested genotype. (3) Treating intercept and slope as new "traits" and perform genomic prediction through ridge regression to predict the intercept and slope for each untested genotype. (4) Predict the phenotypes of the untested genotypes using the predicted intercept and slope and the environmental index value of each environment. Consistent with the RE model, the PEI model starts with identifying key environmental index that best captures the phenotypic variation Total of these predicted statistical models are classified into three major categories: parametric, semi-parametric, and non-parametric (Admas et al, 2024). The parametric statistical models include mixed linear models like genomic best linear unbiased prediction (G-BLUP) (Vanraden, 2008), BayesA (BA) and BayesB (BB) (Meuwissen et al., 2001), BayesC (BC) (George and McCulloch, 1993), Bayesian ridge regression (BRR) (Erbe et al., 2012), and Bayesian LASSO (BL) (Park and Casella, 2008), least absolute shrinkage and selection operator (LASSO) (Usai et al., 2009), ridge regression (RR) (Whittaker et al., 2000), ridge regression best linear unbiased prediction (RR-BLUP) (Meuwissen et al., 2001), and elastic net (EN) (Zou and Hastie, 2005). The semi-parametric method includes the reproducing kernel Hilbert space (RKHS) model and multiple kernel RKHS (MKRKHS) (Gianola et al., 2006). The non-parametric method comprises support vector machine (SVM) (Maenhout et al., 2007), and random forest (RF) (Chen and Ishwaran, 2012), and gradient boosting machine (GBM) (Li et al., 2018). ### Plots
Basic information of phenosData analysis step 1/files/Bd7wMvCgtc39D2pVxQ0x/pages/CyH2xJQs9yWJ1S8BYNav
Search best env indexIndex Search/files/Ivq1FolmiZqdT1S6Ax6w/pages/7aUFmnCMx9m4smGncsXL
Revelance testIndex search/files/YCmgxXavJh4rWvoeRI3n/pages/JjjojIyKxaiBPzzLwvtg