There seems to be a lot of confusion about the influence of genes on phenotypes. I happen to have worked in the field, so I will just give my view on the matter. Geneticists are studying populations and try to find genes that cause the population to be at risk for certain phenotypes. Examples of populations: a family, a country, a race. Examples of phenotypes: severe combined immune deficiency (SCID), alzheimer, atherosclerosis, alcoholism. The best population for a geneticist to use for their studies are isolated populations. There are different kinds of isolated population: * geographic (Pacific island chains, Arabia, Inuits) * cultural (Indian castes, Amish) * historic (Quebec - Sanguenay Lac-Saint-Jean, S. Africans, native Americans) The reason you want to use isolates is that the following are likely: rare alleles are enriched, the population is homogeneous, environmental factors are homogenous and more detailed things I won't go into. It makes it easier to find a correlation. Even with such special population it is important to choose special sampling designs (unrelated case/control, parent/child, relative pairs, nuclear families, extended pedigrees). In isolated populations you have the most power to detect genetic effects when you sample from huge pedigrees, that is why they started the deCODE project in Iceland. The search of these scientists have been succesfull: look for instance at the Finnish disease heritage. There are 35 monogenetic disease that are more prevalent in Finland than in any other population. For more information, this is an excellent website http://www.findis.org/main.php?action=disease [Broken] Does that mean the finns have a sick population? No, diseases that are frequent in other populations are rare in the finns (like cystic fibrosis), it's like a balance. Can you define the finns by those 35 diseases? No, the diseases are distributed regionally. Scientists are most succesfull in finding monogenetic diseases, since their inheritance pattern is easy to detect. Diseases that inherit in a dominant fasion, have a high penetrance and low age at onset are easiest to find. Most importantly they need a very good definition of the phenotype they are looking for. The case is, most genes that cause disease don't follow those rules. Disorders like alcoholism or psychiatric disorders are difficult to define and are heavily influenced by the environment someone grows up in. Only with good matched controls will a study succeed to find a correlation. That is an important point: matched controls, you cannot compare apples with pears. In recent threads I feel that has been the main obstacle. Another important point: the populations. In the discussion before, the populations were very carefully choosen and defined. Finding genes on a mixed genetic background is very difficult. I certainly do believe that certain populations can harbor different genes. The thing that pops to mind most clearly is milk intolerance. Milk tolerance is a mutation, Dr. Leena Peltonen discovered the genetic basis. The following is a nice article http://www.newscientist.com/news/news.jsp?id=ns99991787 but again, this is a very well defined disease and is not influenced by outside factors. I'd be very careful with extrapolating data on poorly defined phenotypes to poorly defined populations.