Premium
Fine Mapping Functional Sites or Regions from Case‐Control Data Using Haplotypes of Multiple Linked SNPs
Author(s) -
Cheng Rong,
Ma Jennie Z.,
Elston Robert C.,
Li Ming D.
Publication year - 2005
Publication title -
annals of human genetics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.537
H-Index - 77
eISSN - 1469-1809
pISSN - 0003-4800
DOI - 10.1046/j.1529-8817.2004.00140.x
Subject(s) - haplotype , single nucleotide polymorphism , contingency table , statistic , test statistic , algorithm , genetics , genetic association , computer science , mathematics , statistics , computational biology , biology , statistical hypothesis testing , allele , genotype , gene
Summary Previously, we reported an algorithm for scanning a large number of tightly linked single nucleotide polymorphisms (SNPs) for LD mapping of functional sites or regions from a family‐based association design. In the present study, we extend our method to a case‐control design. We first use the expectation maximization (EM) algorithm to estimate haplotype frequencies of multiple linked SNPs, and follow this by constructing a contingency table statistic S for LD analysis, based on the estimated haplotype frequencies. An empirical p ‐value is obtained based on the null distribution of the maximum of S ( S *) from a large number (e.g., 1,000 or more) of randomized permutations. The proposed algorithm has been implemented in a computer program in which window searching for functional SNP sites can cover any number of loci without limitation, except that of computer storage. Unlike other programs for a case‐control design that always conduct tests at a fix window width, in our program after setting a maximum size of haplotype window width, for a given maximum window width all possible widths of haplotypes are utilized to find the maximum statistic S * for each locus under investigation. The sensitivity of the proposed algorithm has been examined with simulated and real genotyping datasets. Association analyses indicate that our program is powerful enough to detect most, if not all, functional SNPs simulated in the original model or identified in the original report. Moreover, the program is very flexible and can be used in either regional or genome‐wide scanning for association analysis with SNP markers.