Using character-compatibility matrices in haploblock discovery.

Haplotype blocks (haploblocks) are contiguous stretches of DNA that have not been affected by recombination since the haplotype first evolved. Absence of recombination increases non-random association of alleles at polymorphic sites within the haploblock, and subsequently those polymorphic sites are mutually phylogenetically concordant. These properties enable efficient mapping of disease-associated loci, and enhances our ability to uncover historic demographic events and understand natural selection at discrete genomic regions. Common methods of discovering haploblocks can be divided into two groups: 1) linkage disequilibrium (LD) based methods and 2) methods based on gametic evidence of recombination (e.g. four-gamete test [FGT]). LD-based haploblock discovery methods lack specificity as they can tolerate a degree of recombination-induced noise, which leads to spurious block boundaries. Conversely, the FGT lacks sensitivity towards mutational noise. However, as diploid genomes are spanned by recurrent and back mutations, thereby causing haploblock boundaries to terminate prematurely and thus reducing the number of markers within a haploblock. In contrast to LD-based methods or the FGT, character-compatibility (Co) is another recombination-based method that can allow for removal of mutational noise thus retaining the specificity of the FGT without false-positive haploblock boundaries, as may be discovered by LD-based methods. We have for the first time employed Co-matrices to discover haploblocks in a single population from the 1000 Genomes Project dataset, and compared them to haploblocks discovered with LD-matrices. Our preliminary results suggest that haploblocks discovered using Co-matrices are more phylogenetically concordant and their boundaries terminate earlier than those discovered with LD-matrices, suggesting haploblocks discovered by LD-matrices erroneously extend over recombination points. Given increased specificity and sensitivity of Co compared to the LD-based and the FGT methods respectively, Co may significantly impact our ability to understand the evolutionary history of discrete regions of the human genome, with further possibilities for disease loci mapping.