Interpreting single-step genomic evaluations as mixed effects neural networks of three layers: pedigree, genotypes, and phenotypes

The single-step approach has become the most widely-used methodology for genomic evaluations when only a subset of phenotyped individuals in the pedigree are genotyped, where the genotypes for non-genotyped individuals are imputed based on gene contents of genotyped individuals through their pedigree relationships. We proposed a new method named single-step NN-MM to represent the single-step genomic evaluations as mixed effects neural networks of three sequential layers: pedigree, genotypes, and phenotypes, where the gene contents of non-genotyped individuals are sampled based on pedigree, genotypes, and phenotypes. In simulation analysis, the single-step NN-MM had similar or better prediction performance than the conventional single-step approach. In addition to imputation of genotypes using three sources of information including phenotypes, genotypes, and pedigree, single-step NN-MM provides a more flexible framework to allow nonlinear relationships between genotypes and phenotypes, and individuals being genotyped with different SNP panels. The single-step NN-MM has been implemented in a package called "JWAS".