Testing for signatures of natural selection at molecular genes level

The paper presents the methodology used for detecting the signatures of natural selection at the molecular level from single nucleotide polymorphism data. The results obtained from widely used approach, based on statistical testing departures from neutral evolution model, can be obscured by the presence of alternative hypotheses generating the similar to natural selection results of the tests. These hypotheses include population growth and geographic substructure. Especially for human population these alternatives are of non-negligible importance. In the paper we show how to deal with this problem, both by the analysis of a battery of statistical tests giving indication about the age of the predominant mutations, and by application of non conventional null hypotheses that assume different population scenarios. Since the critical values of the tests are known only for panmicting, constant size population, the second approach demands the intensive computer simulations of coalescence process to obtain analogous critical values for different scenarios used as a null. The methodology with the problem of detecting signatures of natural selection in four genes implicated in human familial cancers has been illustrated.