Close sequence comparisons are sufficient to identify human cis-regulatory elements

摘要: Cross-species DNA sequence comparison is the primary method used to identify functional noncoding elements in human and other large genomes. However, little known about relative merits of evolutionarily close distant comparisons. To address this problem, we identified conserved regions primate, mammalian, more comparisons using a uniform approach (Gumby) that facilitates unbiased assessment impact evolutionary distance on predictive power. We benchmarked computational predictions against previously cis-regulatory at diverse genomic loci also tested numerous extremely human–rodent sequences for transcriptional enhancer activity an vivo assay transgenic mice. Human regulatory were with acceptable sensitivity (53%–80%) true-positive rate (27%–67%) by one five eutherian mammals or six simian primates. More (marsupial, avian, amphibian, fish) failed many empirically defined elements. Our results highlight practical utility comparisons, loss entailed derived intuitive relationship between ancient recent conservation from whole-genome comparative analysis explains most observations empirical benchmarking. Lastly, determined that, addition strength conservation, location and/or density surrounding must be considered selecting candidate enhancers testing embryonic time points.