Genetic dissection of complex diseases

Non mendelian diseases are characterized by complex correspondance between genotypes and phenotypes. Often, the phenotype is determined by the effect of several genes, combined or not with environmental effects, Each susceptibility gene, considered separately, is neither essential nor sufficient to produce the disease. Rather, liability to the disease is a quantitative variable, and the morbid phenotype appears when a liability threshold is reached by the combination of effects of several risk factors. Methods used to map susceptibility genes are reviewed. It is concluded that model dependent linkage studies (e.g. lod score method) are unsuited. Model free linkage methods (affected sib pairs) are more robust but are lacking power to detect risk factors with a small effect and do trot allow a precise localisation. Single Nucleotide Polymorphisms (SNP) based association studies are likely to generate a very large amount of false positive results, given the number of tests realized in a whole genome scan. SNP based association studies are more likely to give firm results when applied to "candidate genes". Use of specific strategies allowing to characterize "candidate genes" is thus strongly advocated. It is suggested that focusing on sub-mendelian entities and intermediate phenotypes (endophenotypes) are a necessary step to define the biological pathways in which susceptibility genes are likely to be found. This approach is illustrated with examples from neuropsychiatric diseases.