A first-generation linkage disequilibrium map of human chromosome 22

Abstract

DNA sequence variants in specific genes or regions of the human genome are responsible for a variety of phenotypes such as disease risk or variable drug response. These variants can be investigated directly, or through their non-random associations with neighbouring markers (called linkage disequilibrium (LD)). Here we report measurement of LD along the complete sequence of human chromosome 22. Duplicate genotyping and analysis of 1,504 markers in Centre d'Etude du Polymorphisme Humain (CEPH) reference families at a median spacing of 15 kilobases (kb) reveals a highly variable pattern of LD along the chromosome, in which extensive regions of nearly complete LD up to 804 kb in length are interspersed with regions of little or no detectable LD. The LD patterns are replicated in a panel of unrelated UK Caucasians. There is a strong correlation between high LD and low recombination frequency in the extant genetic map, suggesting that historical and contemporary recombination rates are similar. This study demonstrates the feasibility of developing genome-wide maps of LD.

Using haplotype blocks to map human complex trait loci

“…Block-like patterns have now been observed in the immunoglobulin cluster on 5q31 [6], within HLA [13] and throughout chromosomes 21 [9] and 22 [14]. In one study [13], these blocks were flanked by precisely localized recombination hotspots, leading to suggestions that PUNCTATE RECOMBINATION could be a general phenomenon underlying block structure [8,10,11].…”

“…Most studies used in silico block definitions that infer block structure and boundaries [6,8,9,14,15]. In general, all of the methods developed aim to identify genomic regions where a few unique haplotypes are distributed widely throughout the population.…”

“…However, each method carries its own set of assumptions and heuristics. Some methods make use of known [9] or inferred haplotypes [6,15], others use pairwise LD measures [6,8] and others a combination of the two [14].…”

“…Specifically, the SNPs currently in public databases and available for genotyping were identified on a small set of chromosomes [16] and are thus not a representative sample of variants in the population. Instead, common alleles are over-represented and rare alleles are under-represented in these samples [8,14,17] (c.f. [9]).…”

“…Thus far, the density of SNPs analysed has ranged from approximately one marker per kilobase of DNA sequence [9,13] to one marker per 15 kb [14]. Although the effect of marker density on inferred block structure is not entirely clear, studies using a denser marker panel have identified more Opinion short blocks [9], whereas studies using sparser marker panels have identified progressively longer blocks [8,14].…”