||New sequencing and genotyping technologies have enabled researchers to generate high density SNP genotype data for mapping populations. In polyploid species, SNP data usually contain a new type of information, the allele dosage, which is not used by current methodologies for linkage analysis and QTL mapping. Here we extend existing methodology to use dosage data on SNPs in an autotetraploid mapping population. Firstly the SNP dosages are inferred from allele intensity ratios using normal mixture models. The steps of the linkage analysis (testing for distorted segregation, clustering of markers, calculation of recombination frequencies and lod scores, ordering of SNPs and inference of parental phase) are extended to use the dosage information. From the SNP order, the parental genotypes and phases and the offspring dosages, the probability of each possible offspring genotype is inferred at a grid of locations along the chromosome. A normal mixture model is then used to model trait values as a function of offspring genotypes and to identify the most likely locations for QTLs.
These methods are applied to analyse a tetraploid potato mapping population of parents and 190 offspring, genotyped using an Infinium 8300 Potato SNP Array. Linkage maps for each of the 12 chromosomes are constructed. The allele intensity ratios are mapped as quantitative traits to check that their position and phase agrees with that of the SNP derived from the intensity ratio. This analysis confirms most SNP positions, and eliminates some problem SNPs to give high-density maps for each chromosome, with between 74 and 152 SNPs mapped and between 100 and 300 further SNPs allocated to approximate bins. Overall 3839 of the 8300 SNPs can be assigned putative genetic locations.
The methodology presented here can be applied to construct high-density linkage maps in any autotetraploid species, and could also be extended to higher autopolyploids.