Complex Diseases Gene's Multilocus Linkage Analysis With Sequential Imputation

Complex diseases is a common disease and a major cause of dying. If we map the complex diseases's gene, the result from large pedigrees is more reliable than from small pedigrees. In disease mapping and marker mapping, multilocus linkage analysis is more reliable. Computationally efficient algorithms for calculating likelihoods are available for large pedigrees with a small number of loci, and small pedigrees with a large number of loci. Elston-Stewart algorithm can solve the large pedigrees with a small number of loci, which is a recursive algorithm that reduces the computational burden of evaluation of the multiple. The algorithm compute a nuclear family at a time from bottom generation to top generation, then peel it. The algorithm scales linearly with the number of pedigree members, but exponentially with the number of loci. Lander-Green algorithm is based on a hidden Markov formation of the pattern of inheritance at several order loci. In contrast to Elston-Stewart algorithm, the Lander-Green algorithm scales exponentially with the number of pedigree members, but linearly with the number of loci. However, for large pedigrees with a large number of loci, especially those that have substantial missing data, the above algorithms can be prohibitive because of the required memory and computing time. In this paper, a method called sequential imputation is proposed to handle problems of this type. Sequential imputation is a Monte Carlo approach that applies the traditional technique of importance sampling in a novel fashion. Based on a fixed value of the parameter, missing data are imputed conditioned on the observed data. The loci are processed one (or a few) at a time to reduce the demand on computational resources. The result is a collection of complete data sets with associated weights. The weights be used to estimate the likelihood of the parameter value used for the imputations. In order to localize BRCA1, polymorphic markers from chromosome 17q12-q21 were screened on families in our series with early-onset. we analysis them using sequential imputation method and Lander-Green algorithm and Elston-Stewart algorithm . Elston-Stewart algorithm had to drop some markers and some allels and Lander-Green algorithm had to drop 22.22% and 55% of pedigree nofounders, whereas sequential imputation was able to use all pedigree members. The results gains of using all pedigree members is more substantial than using partial pedigree members and partial markers.