Method for estimating breeding value by fitting genome with non-additive effect

Abstract

The invention discloses a new method for estimating breeding value by fitting genome with non-additive effect. According to the method, an additive effect prediction model and a non-additive effect prediction model in genome selection are combined into a meta-algorithm of a prediction model,. Compared with a prediction model only fitting an additive effect, the method disclosed by the invention can generally obtain a better prediction effect. The method comprises the following specific steps of acquiring complete genotype information and phenotype information of a single group; randomly dividing a training group and a test group, and performing iterative training on the training group through a hybrid algorithm MixPGV; obtaining an expected additive effect value and an expected non-additive effect value of each SNP site, and performing accumulating to obtain an additive genome estimated breeding value GEBVAdd and a non-additive genome estimated breeding value GEBVNon-Add of an MixPGV prediction model; and accumulating the additive genome estimated breeding value GEBVAdd and the non-additive genome estimated breeding value GEBVNon-Add to obtain a genome estimated breeding value GEBVof the group, and finally calculating a correlation coefficient of the genome estimated breeding value GEBV and a real breeding value to obtain estimation accuracy.

Description

technical field [0001] The invention relates to the field of genome breeding, in particular to a method for estimating the breeding value of a genome fitting non-additive effects. Background technique [0002] Genomic selection (GS) is the better prediction of plant or animal phenotypes for polygenic traits through the use of genome-wide markers. This method exploits linkage disequilibrium (LD) between markers and QTLs to capture small QTL effects. Unlike pedigree-based prediction methods and marker-assisted selection prediction methods based on a small number of markers, genomic selection can predict the breeding value of individuals with unknown phenotypes, and this property allows it to be applied to early selection, which can be greatly improved. Reduce breeding cycles and speed up selection progress. At present, a variety of prediction models have been used to predict the genomic estimated breeding value (GEBV) of traits, such as based on BLUP algorithm (best linear u...

AI Technical Summary

Problems solved by technology

But in hybrids and pure-line breeds, the actual genetic makeup is complex

Images