Analysis of Five Gene Sets in Chimpanzees Suggests Decoupling between the Action of Selection on Protein-Coding and on Noncoding Elements.
Bottom Line: To that effect, we combine human-chimpanzee divergence patterns with polymorphism data obtained from target resequencing 20 central chimpanzees, our closest relatives with largest long-term effective population size.By using the distribution of fitness effect-alpha extension of the McDonald-Kreitman test, we reproduce inferences of rates of evolution previously based only on divergence data on both coding and intronic sequences and also obtain inferences for other classes of genomic elements (untranslated regions, promoters, and conserved noncoding sequences).Our results suggest that 1) the distribution of fitness effect-alpha method successfully helps distinguishing different scenarios of accelerated divergence (adaptation or relaxed selective constraints) and 2) the adaptive history of coding and noncoding sequences within the gene sets analyzed is decoupled.
Affiliation: Departament de Ciències Experimentals i la Salut, Institute of Evolutionary Biology (UPF-CSIC), Universitat Pompeu Fabra, PRBB, Barcelona, Spain.Show MeSH
Related in: MedlinePlus
Mentions: To confirm that adaptive differences between the two reference sets are due to global differentiated patterns between them rather than to outlier genes, we grouped CDS in three different subsets according to the percentile that the CDS of each gene occupies in the distribution of dN/dS values of each pathway and compared genes within categories. In particular, we considered three quantiles (0–25, 25–75, and 75–100; fig. 4, supplementary table S11, Supplementary Material online) and performed DFE-α tests comparing the corresponding subgroups. Estimates of dN/dS per gene were obtained from Serra et al. (2011). These gene dN/dS estimates were averaged for each quantile and compared with the dN/dS values of the corresponding concatenated genes in our dataset. For the Complement pathway, we observed that, within all three quantiles, the higher the dN/dS of the CDS the greater were the α and ωα values. This was not the case for the Actin quantiles, indicating either lack of power or lack of correlation in such a constrained pathway. Thus, we could only carry out the study on the potential contributions of outliers by comparing Complement quantiles against the Actin set taken as a whole. Both α and ωα values were significantly higher than those of the Actin set in Complement’s quantiles 25–75 and 75–100 but not in the 0–25 quantile (P = 0.03; threshold at 0.025). The reciprocal comparison followed the same trend: α and ωα values of the Actin pathway were significantly lower in the two most divergent Complement percentiles but not in the 0–25 quantile (supplementary table S12, Supplementary Material online). The use of 4-fold sites of each gene set as neutral reference did not substantially affect the results (adjusted R2 = 0.995 and 0.92; P value = 0.03 and 0.132, for α and ωα respectively).Fig. 4.—
Affiliation: Departament de Ciències Experimentals i la Salut, Institute of Evolutionary Biology (UPF-CSIC), Universitat Pompeu Fabra, PRBB, Barcelona, Spain.