The AJHG discusses an interesting phenomenon that I noticed too. While replicating a disease marker association the risk allele suddenly reversed which led me believe that there was an error in allele coding. As the authors of this new paper now show these flip-flop association may indeed a true positive association to a non causal variant.
What makes the difference between genetic linkage and association studies? Simply speaking, for linkage you need to inherit a particular marker allele from your parents where it does not matter if a child in another family inherits another allele (pending it shares it with its affected sibling). With association studies this matters.
As we found with the much relaxed linkage strategy so many minor diverse loci, I assume a rather heterogeneous origin of complex diseases. There is no doubt about the importance of genes, but about the sharing of the same genetic abnormality. An (anonymous) position paper on basic Asthma Research Strategy II in Clin Exp All 2006; 36: 1326 says
The average size of effect on asthma and related traits from common SNPs is small. For instance, seven common SNPs in the IL13 gene jointly accounted for only 0,5% of the variance of total IgE … With a heritability of circa 60% for total IgE this implies that hundreds of genes, each with small effects, may be involved in IgE regulation.
Families presenting with a complex system disease will all have unique patterns how they arrive at the same clinical endpoint. Alpha-delta-gamma asthma, theta-kappa-jota schizophrenia or $%&# diabetes – are they combining lets say 3000 variations in 300 genes of 30 metabolic-signalling pathways to 1 disease of variable onset, severity and prognosis? Yea, yea.