Positive and negative gene association data cluster in hot and cold spots around the world

If there is one factor common to genetic association studies it is their heterogeneity Bertram and Tanzi 2004 Shi et al, 2008 Dozen of confusing positive and negative studies blur the picture and suggest that many are related to artefact or insufficient statistical power. However, while at a global level these results are contentious, at a local level they can be quite homogenous, tending to cluster in positive or negative association hot and coldspots around the world.These examples are taken from data from the Alzgene and Szgene databases and are posted to illustrate this phenomen.
This clustering could relate to several possibilities. Firstly, all polymorphisms/mutations have a single ancestral origin in a particular place at a particular time. For example a presenilin mutation has been tracked back to a family in Southern Italy prior to the 17th century Bruni at al, 2010 Their spread around the world will depend on their success (natural selection), their age and the migratory patterns of their bearers. Conquests, forced and voluntary expatriation will influence their distribution. Secondly, environmental factors may influence the risk promoting effects of many genes. For example many Alzheimer's disease related genes are related to cholesterol and lipoprotein function, and many related risk factors See risk factors (high cholesterol, saturated fat consumption) and protective factors (fish diets, polyunsaturated fat consumption, Mediterranean diet) may influence whether or not these genes are risk promoting. High fat consumption and APOE polymorphisms are a dangerous mix, but perhaps APOE is less risk promoting in countries with a high fish consumption (see negative APOE data from Japan). Thirdly, certain genes may also be related to infectious agents Carter 2009  Carter 2010  Prasad et al, 2010 whose distribution varies around the globe. Each of these could influence positive and negative association data, which at a local level do not seem to be so heterogeneous.