In the case of the human lineage, where functional elements may have zero expected substitutions, acceleration tests can reach genome-wide Caspase inhibitor significance even when there are only a few human-specific substitutions (i.e. not many
more than expected under a neutral model). Hence, tests for acceleration can be more powerful than those for selection. Nonetheless, many accelerated regions do show signatures of positive selection (see below). The goal of a test for accelerated evolution is to determine if the rate of DNA substitutions is faster than expected in a lineage of interest. This lineage can be a single branch (e.g. human since divergence from chimp), a clade (e.g. great apes), or an
extinct species (e.g. ancestor of all primates). A variety of tests have been proposed, including ones that estimate substitutions via models of molecular evolution [23 and 54] and ones that compare parsimony-inferred counts of substitutions [21 and 22]. Some tests make use of the phylogenetic relationships between species to derive expected numbers of substitutions in the lineage of interest, while others directly compare sister species. Regardless of these distinctions, the idea is to determine whether the data in a multiple sequence alignment is more consistent with lineage-specific acceleration versus the expected rate of substitutions. This cross-species approach is related to, but distinct from, methods that employ polymorphism data to identify selection within a species [55]. The data used to identify selleck chemical accelerated regions are aligned DNA sequences from multiple species with a phylogenetic tree, which is either known a priori or computed from the sequence data. There are also specialized comparative genomics methods for identifying slow and fast evolving proteins [16 and 56] or RNA genes [57], which use alignments of codons, amino acids, or structured RNA, as well as methods based
on loss and gain of regulatory motifs (Siepel and Arbiza, in this issue) [58]. These are powerful approaches for studying specific small subsets of the genome, Branched chain aminotransferase but DNA-based methods are needed for unbiased, genome-wide scans. Whole genomes can in principle be analyzed for lineage-specific acceleration one base pair (bp) at a time, although this approach has very low power compared to testing windows 100 bp or larger [54]. To focus on functional windows of the genome, analyses have typically used evolutionarily conserved elements. Because acceleration on the lineage of interest may prevent a region from being classified as conserved, this lineage should be removed from the alignment before generating the conserved elements [4•]. Acceleration tests can also be applied to neutral regions to detect gain-of-function events, provided the regions are long enough to have sufficient power.