Simplistically speaking linkage studies are geared for relatively rare alleles with large effects within families (that may be of little effect in the population), whereas association studies are designed to detect common genetic effects that have smaller effects. In the case of rare monogenic disease, multiple rare variants at linked loci (allelic heterogeneity) seem to be the rule not the exception.42 For common polygenic disease and quantitative traits this question is still unanswered – there are examples for both common43 and rare alleles,44 and theoretical and empirical studies suggest a role for both rare and common variants.45This is a very important (and under-appreciated) issue. People have a genetic predisposition to be fat or thin; we are certain of as much. But what kinds of genes are responsible for this?
The modest LOD scores we observed for BMI despite the large study sample likely
reflects the heterogeneity of our study populations, and may suggest that there are relatively few common loci with strong effects for BMI across these populations.
Under the "Common loci" explanation, in a population there is a limited number of genes which affect one's weight: if you get the "fat" genes you are likely to be fat, if you get the "thin" ones, you are likely to be thin.
An association study boils down to looking at people's genes and trying to correlate them with the trait of interest, e.g. their body mass index.
But, to discover such an association, people must be "fat" or "thin" because of the same genes. These genes must thus be "common" in the population.
If on the other hand, people are thin or fat because of family-specific genes, i.e. uncommon genes that are limited to families and are "uncommon" in the population, an association study can't detect them: it will notice that there are "fat" and "thin" people, but won't be able to find a common genetic pattern distinguishing the two.
A family-based linkage study, on the other hand, looks at the genes inherited by children from a parent. If, e.g. a fat father and thin mother have a fat daughter and a thin son, it pays off to see which genes were inherited by the daughter from the father: chances are, some of them may be the rare family-specific genes responsibly for her weight.
What this study has found is that rare rather than common loci are responsible for body mass index. In other words, people have a predisposition to be fat or thin mainly not because of genes that abound in the population, but because of rare genes common in a family.
If you have been paying attention in the last few years, you'd have noticed that genetic effects of discovered loci in association studies for quantitative traits have been rather underwhelming, explaining a very small (rarely more than 10%, usually less) percentage of the variation.
There are two reasons for this: first, many genes are responsible for each trait, but, more importantly, that a lot (in my opinion most) of the variation for quantitative traits is due to rare variants that can't be discovered by association studies.
The implications of this realization are manifold, but here are two:
- Companies such as 23andme and decodeme who offer personal genome scans aren't likely to offer any really interesting information to consumers any time soon. On the one hand, for legal reasons, they are unlikely to dabble into Mendelian traits that have big and dramatic effects on consumers' health. On the other, they can't figure out which particular family-inherited genes have a major impact on their customers' health (unless multiple family members get tested at several $100 each). Thus, they have to make do with the common alleles discovered in association studies that have minor effects. No wonder prices are dropping.
- Real progress will only come about with more developmental and functional studies, i.e. studies that actually look at what genes do in the body. Note, that in an association study, you don't really need to know what effect a particular allele has: if you discover a statistically significant correlation between its presence and a phenotype, you're done. But, to make progress, we have to understand how phenotypes are produced. Then, instead of estimating that a person will be obese because he carries particular gene found to have an association with obesity in population samples, we will be able to tell what effects the genes will actually have in his body. Naturally, this is easier said than done.
European Journal of Human Genetics doi: 10.1038/ejhg.2008.152
Genome-wide linkage screen for stature and body mass index in 3.032 families: evidence for sex- and population-specific genetic effects
Sampo Sammalisto et al.
Stature (adult body height) and body mass index (BMI) have a strong genetic component explaining observed variation in human populations; however, identifying those genetic components has been extremely challenging. It seems obvious that sample size is a critical determinant for successful identification of quantitative trait loci (QTL) that underlie the genetic architecture of these polygenic traits. The inherent shared environment and known genetic relationships in family studies provide clear advantages for gene mapping over studies utilizing unrelated individuals. To these ends, we combined the genotype and phenotype data from four previously performed family-based genome-wide screens resulting in a sample of 9.371 individuals from 3.032 African-American and European-American families and performed variance-components linkage analyses for stature and BMI. To our knowledge, this study
represents the single largest family-based genome-wide linkage scan published for stature and BMI to date. This large study sample allowed us to pursue population- and sex-specific analyses as well. For stature, we found evidence for linkage in previously reported loci on 11q23, 12q12, 15q25 and 18q23, as well as 15q26 and 19q13, which have not been linked to stature previously. For BMI, we found evidence for two loci: one on 7q35 and another on 11q22, both of which have been previously linked to BMI in multiple populations. Our results show both the benefit of (1) combining data to maximize the sample size and (2) minimizing heterogeneity by analyzing subgroups where within-group variation can be reduced and suggest that the latter may be a more successful approach in genetic mapping.