Use of partial least squares regression to predict single nucleotide polymorphism marker genotypes when some animals are genotyped with a low-density panel

Abstract

High-density single nucleotide polymorphism (SNP) platforms are currently used in genomic selection (GS) programs to enhance the selection response. However, the genotyping of a large number of animals with high-throughput platforms is rather expensive and may represent a constraint for a large-scale implementation of GS. The use of low-density marker (LDM) platforms could overcome this problem, but different SNP chips may be required for each trait and/or breed. In this study, a strategy of imputation independent from trait and breed is proposed. A simulated population of 5865 individuals with a genome of 6000 SNP equally distributed on six chromosomes was considered. First, reference and prediction populations were generated by mimicking highand low-density SNP platforms, respectively. Then, the partial least squares regression (PLSR) technique was applied to reconstruct the missing SNP in the low-density chip. The proportion of SNP correctly reconstructed by the PLSR method ranged from 0.78 to 0.97 when 90% and 50%, respectively, of genotypes were predicted. Moreover, data sets consisting of a mixture of actual and PLSR-predicted SNP or only actual SNP were used to predict genomic breeding values (GEBVs). Correlations between GEBV and true breeding values varied from 0.74 to 0.76, respectively. The results of the study indicate that the PLSR technique can be considered a reliable computational strategy for predicting SNP genotypes in an LDM platform with reasonable accuracy.

Use of partial least squares regression to impute SNP genotypes in Italian Cattle breeds

“…PLSR is a multivariate statistical covariance-based technique that is able to predict a response matrix Y ( n × p ) from a predictor matrix X ( n × m ) and to describe the common structure of the two matrices [18]. In both X and Y , n represents the number of animals involved, m is the number of SNPs in the LDP and p is the number of SNPs to be imputed.…”

“…PLSR allows for the identification of underlying variables (known as latent factors) which are linear combinations of the explanatory variables X , that best model Y . Dimauro et al [18] demonstrated that the accuracy of PLSR prediction increases with the number of latent factors approaching the number of SNPs to be predicted (the columns of Y ). The maximum number of latent factors depends on the size of X , which has a lower number of columns than Y .…”

“…PLSR is a multivariate statistical technique particularly useful in genomic studies in which a great number of variables are involved. It can overcome the strong collinearity between SNP variables in X or Y and, at the same time, maximize correlations between Y and X variables [18,19]. A more detailed description of the PLSR imputation method can be found in Dimauro et al [18].…”

“…A method that uses the Partial Least Squares Regression (PLSR) technique to impute SNP genotypes was proposed recently [18]. It was tested on a simulated genome consisting of 6000 SNPs equally distributed on six chromosomes and a data set of 5865 individuals (TP = 4665 and PP = 1200).…”

“…In the latter case, the accuracy of direct genomic values (DGV) dropped from 0.77 to 0.74. Furthermore, Dimauro et al [18] highlighted that, with a fixed percentage (50%) of SNPs to be predicted, imputation accuracies slowly decreased from 98% with TP = 5000, to 87% with TP = 1000 and to 69% with TP = 600. PLSR requires only genotype data, and other data, such as pedigree relationships, is not needed.…”