Comparison of two QTL mapping approaches based on Bayesian inference using high-dense SNPs markers

To compare different QTL mapping methods, a population with genotypic and phenotypic data was simulated. In Bayesian approach, all information of markers can be used along with combination of distributions of SNP markers. It is assumed that most of the markers (95%) have minor effects and a few numbers of markers (5%) exert major effects. The simulated population included a basic population of 1020 non-relative cattle that was continuously crossed for 4 generations to make disequilibrium linkage among QTL position and markers. In all generations, 20 bulls were mated with 1,000 cows and each cow produced only one offspring. Whole tree family included 4100 head of livestock. Genotype of 10000 SNPs on 5 chromosomes at equal distance (0.05 cM) in the total population was simulated. The length of each chromosome was 100 cM. Simulated trait was milk production. Progeny of the first to third generation had record but the basic population and fourth generation of offspring did not have any record. Therefore, from the total population of 4100 heads, 3000 cattle had record. Two different models, Bayz A and Bayz B, were used to analyze QTL linked to the SNP markers. Analysis was conducted by BAYZ software. SNPs with more than 0.6 effect or Bayes factor (BF) greater than 5.5 were considered as QTL. The resultant analysis of two models of BAYZ A and BAYZ B were 7 and 9 QTL locations on 5 chromosomes, respectively. QTL position identified by BAYZ B method was matched on simulated location, but showed a false positive on chromosome 4. QTL positions identified by BAYZ A method were located near by the simulated positions, but with many false positive points.