Bioinformatic Studies on Selection in Protein-Coding- and RNA-Specifying Genes

During my dissertation I used a combination of comparative and phylogenetic approaches to test for selection on genic regions. Specifically, I tested for mutational and environmental robustness in Drosophila pre-miRNAs, as well as selection on synonymomous and nonsynonymous sites in mammalian protein coding genes. Contrary to previous claims of selection for mutational robustness in mammalian pre-miRNAs, I demonstrate that mutational robustness evolved neutrally in Drosophila pre-miRNas. Furthermore, I show that mutational robustness did not evolve as a byproduct of selection for environmental robustness. In Chapter 3, I identify orthologous processed pseudogenes and use them to test for selection on synonymous sites. By estimating the rate of substitution at synonymous sites in genes (d_S), and corresponding sites in pseudogenes (d_ψ), I demonstrate that only about ~8% of synonymous changes could possibly be under selection. This is in stark contrast to the ~30% previously claimed. However, I show that both deviations from neutrality can be caused by an increase in divergence between the sequences examined, in combination, with a difference in nucleotide composition between genes and pseudogenes. In the last part of my dissertation, I estimate selection on nonsynonymous sites in mammals (d_N/d_S), and examine the previously claimed positive correlation between generation time and selection. To ensure the correct estimation of selection, I examine the effects of sequencing errors and alignment quality. After accounting for phylogenetic independence, I find no correlation between generation time and selection. Furthermore, I find a significant decrease in the efficiency of selection in monkeys, after the simian and prosimian split.