Sawyer Stanley A, Parsch John, Zhang Zhi, Hartl Daniel L
Department of Mathematics, Washington University, St. Louis, MO 63130, USA.
Proc Natl Acad Sci U S A. 2007 Apr 17;104(16):6504-10. doi: 10.1073/pnas.0701572104. Epub 2007 Apr 4.
We have estimated the selective effects of amino acid replacements in natural populations by comparing levels of polymorphism in 91 genes in African populations of Drosophila melanogaster with their divergence from Drosophila simulans. The genes include about equal numbers whose level of expression in adults is greater in males, greater in females, or approximately equal in the sexes. Markov chain Monte Carlo methods were used to sample key parameters in the stationary distribution of polymorphism and divergence in a model in which the selective effect of each nonsynonymous mutation is regarded as a random sample from some underlying normal distribution whose mean may differ from one gene to the next. Our analysis suggests that approximately 95% of all nonsynonymous mutations that could contribute to polymorphism or divergence are deleterious, and that the average proportion of deleterious amino acid polymorphisms in samples is approximately 70%. On the other hand, approximately 95% of fixed differences between species are positively selected, although the scaled selection coefficient (N(e)s) is very small. We estimate that approximately 46% of amino acid replacements have N(e)s < 2, approximately 84% have N(e)s < 4, and approximately 99% have N(e)s < 7. Although positive selection among amino acid differences between species seems pervasive, most of the selective effects could be regarded as nearly neutral. There are significant differences in selection between sex-biased and unbiased genes, which relate primarily to the mean of the distributions of mutational effects and the fraction of slightly deleterious and weakly beneficial mutations that are fixed.
我们通过比较非洲黑腹果蝇群体中91个基因的多态性水平及其与拟果蝇的差异,估算了自然群体中氨基酸替换的选择效应。这些基因中,在成虫中表达水平在雄性中更高、在雌性中更高或两性中大致相等的数量大致相同。马尔可夫链蒙特卡罗方法被用于在一个模型中对多态性和差异的平稳分布中的关键参数进行采样,在该模型中,每个非同义突变的选择效应被视为从某个潜在正态分布中随机抽取的样本,其均值可能因基因而异。我们的分析表明,所有可能导致多态性或差异的非同义突变中,约95%是有害的,样本中有害氨基酸多态性的平均比例约为70%。另一方面,物种间约95%的固定差异是正选择的,尽管标度选择系数(N(e)s)非常小。我们估计,约46%的氨基酸替换的N(e)s < 2,约84%的N(e)s < 4,约99%的N(e)s < 7。尽管物种间氨基酸差异的正选择似乎普遍存在,但大多数选择效应可被视为近乎中性。性别偏向基因和非偏向基因之间在选择上存在显著差异,这主要与突变效应分布的均值以及固定的轻度有害和弱有益突变的比例有关。