AGHmatrix (version 2.1.4)

Gmatrix: Construction of Relationship Matrix G

Description

Given a matrix (individual x markers), a method, a missing value, and a maf threshold, return a additive or non-additive relationship matrix. For diploids, the methods "Yang" and "VanRaden" for additive relationship matrices, and "Su" and "Vitezica" for non-additive relationship matrices are implemented. For autopolyploids, the method "VanRaden" for additive relationship, method "Slater" for full-autopolyploid model including non-additive effects, and pseudo-diploid parametrization are implemented. Weights are implemented for "VanRaden" method as described in Liu (2020).

Usage

Gmatrix(
  SNPmatrix = NULL,
  method = "VanRaden",
  missingValue = -9,
  maf = 0,
  thresh.missing = 0.5,
  verify.posdef = FALSE,
  ploidy = 2,
  pseudo.diploid = FALSE,
  integer = TRUE,
  ratio = FALSE,
  impute.method = "mean",
  rmv.mono = FALSE,
  thresh.htzy = 0,
  ratio.check = TRUE,
  weights = NULL,
  ploidy.correction = FALSE,
  ASV = FALSE
)

Value

Matrix with the marker-bases relationships between the individuals

Arguments

SNPmatrix

matrix (n x m), where n is is individual names and m is marker names (coded inside the matrix as 0, 1, 2, ..., ploidy, and, missingValue).

method

"Yang" or "VanRaden" for marker-based additive relationship matrix. "Su" or "Vitezica" for marker-based dominance relationship matrix. "Slater" for full-autopolyploid model including non-additive effects. "Endelman" for autotetraploid dominant (digentic) relationship matrix. "MarkersMatrix" for a matrix with the amount of shared markers between individuals (3). Default is "VanRaden", for autopolyploids will be computed a scaled product (similar to Covarrubias-Pazaran, 2006).

missingValue

missing value in data. Default=-9.

maf

minimum allele frequency accepted to each marker. Default=0.

thresh.missing

threshold on missing data, SNPs below of this frequency value will be maintained, if equal to 1, no threshold and imputation is considered. Default = 0.50.

verify.posdef

verify if the resulting matrix is positive-definite. Default=FALSE.

ploidy

data ploidy (an even number between 2 and 20). Default=2.

pseudo.diploid

if TRUE, uses pseudodiploid parametrization of Slater (2016).

integer

if FALSE, not check for integer numbers. Default=TRUE.

ratio

if TRUE, molecular data are considered ratios and its computed the scaled product of the matrix (as in "VanRaden" method).

impute.method

"mean" to impute the missing data by the mean per marker, "mode" to impute the missing data by the mode per marker, "global.mean" to impute the missing data by the mean across all markers, "global.mode" to impute the missing data my the mode across all marker. Default = "mean".

rmv.mono

if monomorphic markers should be removed. Default=FALSE.

thresh.htzy

threshold heterozigosity, remove SNPs below this threshold. Default=0.

ratio.check

if TRUE, run Mcheck with ratio data.

weights

vector with weights for each marker. Only works if method="VanRaden". Default is a vector of 1's (equal weight).

ploidy.correction

It sets the denominator (correction) of the crossprod. Used only when ploidy > 2 for "VanRaden" and ratio models. If TRUE, it uses the sum of "Ploidy" times "Frequency" times "(1-Frequency)" of each marker as method 1 in VanRaden 2008 and Endelman (2018). When ratio=TRUE, it uses "1/Ploidy" times "Frequency" times "(1-Frequency)". If FALSE, it uses the sum of the sampling variance of each marker. Default = FALSE.

ASV

if TRUE, transform matrix into average semivariance (ASV) equivalent (K = K / (trace(K) / (nrow(K)-1))). Details formula 2 of Fieldmann et al. (2022). Default = FALSE.

Author

Rodrigo R Amadeu , Marcio Resende Jr, Letícia AC Lara, Ivone Oliveira, and Felipe V Ferrao

References

Covarrubias-Pazaran, G. 2016. Genome assisted prediction of quantitative traits using the R package sommer. PLoS ONE 11(6):1-15.

Endelman, JB, et al., 2018. Genetic variance partitioning and genome-wide prediction with allele dosage information in autotetraploid potato. Genetics, 209(1) pp. 77-87.

Feldmann MJ, et al. 2022. Average semivariance directly yields accurate estimates of the genomic variance in complex trait analyses. G3 (Bethesda), 12(6).

Liu, A, et al. 2020. Weighted single-step genomic best linear unbiased prediction integrating variants selected from sequencing data by association and bioinformatics analyses. Genet Sel Evol 52, 48.

Slater, AT, et al. 2016. Improving genetic gain with genomic selection in autotetraploid potato. The Plant Genome 9(3), pp.1-15.

Su, G, et al. 2012. Estimating additive and non-additive genetic variances and predicting genetic merits using genome-wide dense single nucleotide polymorphism markers. PloS one, 7(9), p.e45293.

VanRaden, PM, 2008. Efficient methods to compute genomic predictions. Journal of dairy science, 91(11), pp.4414-4423.

Vitezica, ZG, et al. 2013. On the additive and dominant variance and covariance of individuals within the genomic selection scope. Genetics, 195(4), pp.1223-1230.

Yang, J, et al. 2010. Common SNPs explain a large proportion of the heritability for human height. Nature genetics, 42(7), pp.565-569.

Examples

if (FALSE) {
## Diploid Example
data(snp.pine)
#Verifying if data is coded as 0,1,2 and missing value.
str(snp.pine)
#Build G matrices
Gmatrix.Yang <- Gmatrix(snp.pine, method="Yang", missingValue=-9, maf=0.05)
Gmatrix.VanRaden <- Gmatrix(snp.pine, method="VanRaden", missingValue=-9, maf=0.05)
Gmatrix.Su <- Gmatrix(snp.pine, method="Su", missingValue=-9, maf=0.05)
Gmatrix.Vitezica <- Gmatrix(snp.pine, method="Vitezica", missingValue=-9, maf=0.05)

## Autetraploid example
data(snp.sol)
#Build G matrices
Gmatrix.VanRaden <- Gmatrix(snp.sol, method="VanRaden", ploidy=4)
Gmatrix.Endelman <- Gmatrix(snp.sol, method="Endelman", ploidy=4) 
Gmatrix.Slater <- Gmatrix(snp.sol, method="Slater", ploidy=4)
Gmatrix.Pseudodiploid <- Gmatrix(snp.sol, method="VanRaden", ploidy=4, pseudo.diploid=TRUE) 

#Build G matrix with weights
Gmatrix.weighted <- Gmatrix(snp.sol, method="VanRaden", weights = runif(3895,0.001,0.1), ploidy=4)
}