Methods For Haplotype Reconstruction Based On Linkage Disequilibrium In Natural Population

Haplotype analysis is becoming increasingly popular for mapping gene underlyingcomplex traits because it is more informative and cost-effective than SNP analysis. Withhaplotype information, target gene could be precisely located in certain genomic region. Thisapproach has been successfully applied in positional cloning for Cystic fibrosis andHuntington disease. Generally genotyping methods can only detect the genotype ofindividuals while phase information is unavailable. Although haplotype could be obtainedexperimentally, these methods are limited by either considerable cost or complicatedtechniques, these defects make them unrealistic to be employed in large-scale positionalcloning. Consequently statistical methods are applied to infer haplotypes when genotypeinformation has been obtained.Three statistical methods, Clark, EM and Phase algorithms, are widely used for haplotype reconstruction. Most researchers believe that higher linkage disequilibrium (LD) level may lead to higher accuracy of haplotype inference. However, the comparison of different methods' efficiency under a given LD value is absent. In our study a population is constructed to evolve over generations by computer simulation. The performance of the three methods is analyzed using LD over generations. Results show that phase and EM algorithms are more accurate than Clark's for haplotype frequency estimation and individual haplotype inference. A method is proposed to formulate the relationship between accuracy of phase, EM algorithm and LD in Hardy-Weinberg populations. This method is proved consistent well with the performance of the phase and EM algorithms at different LD levels.