An Effective Algorithm for SNP Haplotype Block Inference

Abstract

Recently, it has been shown that there should exist a block-like structure in human genome, and thus only limited haplotype diversity can be obtained. In this paper, we propose a fixed-diversity strategy to find out the suitable block diversity and block boundaries. The diversity value in one block is defined as d 1

åni

1 xi

2, where xi denotes the frequency ratio of the ith kind of haplotype within the block and n denotes the number of distinct types of haplotype. We figure out that once a putative block stretches across the primary block boundaries, the diversity will increase rapidly. And the secondary block boundary effects occur when two types are merged or split into different types. The threshold in our algorithm is decided by the two detections of the primary and secondary block boundary effects. We obtain a reasonable diversity of chromosome 21 SNP data with our algorithm. Our partition result shows highly concurrence property to the haplotype data downloaded from NCBI website.