Most commonly, linkage disequilibrium (LD) is analysed per pair of loci, giving one LD-measure per pair. These are, however, not independent. I have started looking at ways of analysing multilocus LD to extract information on the origin and evolution of the polymorphisms.
One critical confounding factor is that population inhomogeneities, i.e. subpopulations with different polymorphism frequencies, give spurious LD. Since analyses multilocus LD are more sensitive toward uncertainties and deviations from the naive model, an ability to either rule out or correct for population inhomogeneity is needed.
This project never passed the thinking stage.