Four methods to detect positive selection from the genetic data of modern human populations

Sabeti, P. C., S. F. Schaffner, et al. (2006). “Positive Natural Selection in the Human Lineage.” Science 312(5780): 1614-1620.

As molecular genetics technology advances, it has become much easier to analyze genes or regions of your interest to see if they show evidence of positive selection.  There are many articles published in last 10 years or so that address positive selection in human populations.  Many different methods are used, but in their articles, Sabeti et al. reviewed major methods to identify genetic signature of positive selection.  This is a quick note on four methods to detect positive selection from the genetic data of modern human populations.

1. Differences between populations  Natural selection tends to be localized, so large allele frequency differences and large Fst should be observed between geographically distant populations.
2. High frequency of derived alleles  If derived allele is advantageous and the effect of selection is large, derived allele frequency increase quickly.
3. Reduction in genetic diversity   Frequency of allele at the loci linked to the positively selected allele increases with frequency of positively selected allele.
4. Long-range haplotypes   Recombination usually breaks the link between these loci, but if the effect of selection is strong, the linkage between these loci extends for very long.

There are three important things to consider.

1. Demographic events, such as bottleneck, expansion, and population subdivision, leave similar genetic signatures, so first, we should consider how demographic events affected the genetic variation.  Also, we should examine, if we can observe similar pattern in different genes or regions, which are not positively selected.
2. If positive selection had small effects on genetic variation, these four methods will not detect the signature, so there are a lot of positively selected genes that we can identify with these methods.
3. If genetic data from publically available genomic database, such as HapMap and CEPH-Human Genome Diversity Project, effects of ascertainment bias need to be considered (problems associated with ascertainment bias will be discussed in the next post).

Advertisement