Rule-based induction method for haplotype comparison and identification of candidate disease loci

One of the most important goals in biomedical research is to identify genes that predispose humans to diseases, such as cancer. To facilitate the identification of these genes, a number of genome-wide approaches have been suggested, such as genetic linkage and genome-wide association (GWA) methods [1]. The linkage methods have revealed several high penetrance disease susceptibility loci [1], whereas GWA studies have been useful in the 'common disease - common allele' model [2]. However, neither approach is well suited to tackle moderate penetrance susceptibility because such a condition rarely results in large pedigrees, with few or no phenocopies, convenient for linkage analysis. As the GWA approaches cannot detect these presumably rare alleles, there is clearly a need for methods that are able to identify loci where such variants could be located. Evolutionarily recent, and thus rare, mutations are usually conveyed in a pedigree by a shared haplotype. Therefore, detection of such haplotypes can lead to the identification of rare or moderate penetrance variants behind disease susceptibility.We introduce here Haplous, a novel computational approach that uses phased genotype data, such as genome-wide SNPs, to identify and prioritize genomic regions likely to be inherited from a common ancestor. The central idea of our approach is to use haplotypes, instead of single alleles, and rank them based on expert-defined rules that determine the haplotypes shared in heterozygous and homozygous forms. As the identification of haplotypes has been recognized as useful for revealing disease predisposing genes, several haplotype association methods have been developed [3-10]. These methods include detection of haplotype diversity and statistical association tests. Haplotypes can be detected with fixed or variable length sliding window [7,11,12], haplotype blocks [13], haplotype clustering [9], a cladistic approach [10] or considering non-contiguous haplotypes [5,8]. Some haplotype