Family-based linkage and association-based case-control studies are both required to identify the genes involved in a disease such as tuberculosis, where several genes are likely to be involved in the aetiology of the disease. The two study designs are complementary because they each have their advantages and limitations. Linkage studies can now be used to screen the entire human genome to locate regions containing disease susceptibility genes (Davies et al 1994). This comprehensive and systematic approach should identify any gene that exerts a major effect on disease susceptibility, but because it has relatively low power it will fail to detect genes that exert only a moderate effect on disease risk. Risch & Merikangas (1996) discuss the difference in power between linkage and association, and the effects which allele frequency and additional disease risk attributable to possession of that allele have upon the sample size required. For example, they show that if a disease susceptibility allele has a frequency of 0.5 and exerts a twofold risk of disease, then to have an 80% chance of detecting this effect one would require 2498 sib pairs by linkage but only 340 cases by association. Association studies are therefore useful to detect genes that exert a moderate disease susceptibility effect and that would be missed by linkage. However, as association is only detectable over short genetic distances (generally < 1 cM) it is not possible to conduct a whole genome screen using this approach, and association studies are currently limited to the screening of candidate regions. The limited regions over which association is detectable can also be used to advantage. Once a linkage has identified a region segregating with disease, association can be used in the fine mapping of the region to localize the disease susceptibility gene, as was recently carried out for haemochromatosis (Feder et al 1996). We will first discuss the methodology of a genome screen and the progress that has been made in tuberculosis, and then describe association-based candidate gene studies.
Was this article helpful?