A novel linkage-disequilibrium corrected genomic relationship matrix for SNP-heritability estimation and genomic prediction

摘要: Single nucleotide polymorphism (SNP)-heritability estimation is an important topic in several research fields, including animal, plant and human genetics, as well ecology. Linear mixed model of SNP-heritability uses the structures genomic relationships between individuals, which constructed from genome-wide sets SNP-markers that are generally weighted equally their contributions. Proposed methods to handle dependence SNPs include, “thinning” marker set by linkage disequilibrium (LD)-pruning, use haplotype-tagging SNPs, LD-weighting SNP-contributions. For improved estimation, we propose a new conceptual framework for relationship matrix, Mahalanobis distance-based LD-correction used linear SNP-heritability. The superiority presented method illustrated compared mixed-model analyses using VanRaden matrix GCTA employing (as implemented LDAK software) simulated (using real human, rice cattle genotypes) (maize, mice) datasets. Despite computational difficulties, our results suggest proposed one can improve accuracy estimates datasets with high LD.