a matrix or data frame in which the genotypes of all SNPs are coded either by 0, 1 and 2, or by 1, 2 and 3, or by "AA", "AB" and "BB". Missing values are allowed. In the latter coding not only NA, but also "NN" is allowed for specifying missing values. Using the former two codings it is assumed that the smallest value codes the homozygous reference genotype, the second value the heterozygous genotype, and the largest value the homozygous variant genotype. For the third coding, see refAA.

should each SNP be coded by two dummy variables from which one codes for a recessive, and the other for a dominant effect? If TRUE, then the first binary variable is set to 1 if the SNP is of the heterozygous or the homozygous variant genotype, and the second dummy variable is set to 1 if the SNP is of the homozygous variant genotype. If FALSE, three dummy variables are used and each of the three genotypes of a SNP is coded by one of these binary variables.

codes "AA" always for the homozygous reference gentoype? Only considered if the SNPs are coded by "AA", "AB" and "BB". If FALSE, it is evaluated SNPwise whether "AA" or "BB" occurs more often, and the more frequently occuring value is assumed to be the homozygous reference genotype.

does each column of mat correspond to a SNP (and each row to an observation)? If FALSE, it is assumed that each row represents a SNP, and each column an observation.

a non-negative number. If a dummy variable contains monomorph or less values that differ from the more frequent value of this variable, then the variable is removed from the data set.