How does PLINK work?

Question

[L_to_the_m]

Hi, I would like to know how a logistic regression in a GWAS works in detail. I have an example dataset, with SampleIDs, Genotypes (GT) in dosage formt and binary phenotypes. Now, how would the logistic regression be performed if I only want to work with the allele counts/frequencies and not the genotypes? Or is the logistic regression always performed on the genotypes? Does PLINK also calculate it on the genotypes? Does it make sense to calculate a 2x3/2x2 contingency table first? Or is PLINK calculating the contingency tables and then does a logistic regression?

SNP1

SampleID                       GT  Pheno
Sample1                         1   1
Sample2                         0   0
Sample3                         2   0
Sample4                         0   1
Sample5                         1   1