Linkage disequilibrium measures for fine-scale mapping: a comparison

We investigate properties of simple linkage disequilibrium mapping for five measures in the presence of mutation at the marker and/or the disease locus and of initial incomplete linkage disequilibrium. In contrast to the stimulation approach that Devlin and Risch used, we calculate the expected values of various linkage disequilibrium measures under different assumptions based on a framework for linkage disequilibrium mapping. These expected values clearly demonstrate the expected performance of these measures. We find that the impact of marker mutation on their performance depends on the magnitude of the mutation relative to the proximity of the marker (i.e. recombination fraction between the marker and the disease locus). In the presence of recurrent mutation at the marker and/or disease locus, the performance of all measures, including the robust one, depends on the marker allele frequency. The initial incomplete linkage disequilibrium could render all measures useless. These expected values also show clearly why in Devlin and Risch's simulation some measures performed very badly under certain circumstances.