An ancient evolutionary origin of genes associated with human genetic diseases

Abstract

Several thousand genes in the human genome have been linked to a heritable genetic disease. The majority of these appear to be nonessential genes (i.e., are not embryonically lethal when inactivated), and one could therefore speculate that they are late additions in the evolutionary lineage toward humans. Contrary to this expectation, we find that they are in fact significantly overrepresented among the genes that have emerged during the early evolution of the metazoa. Using a phylostratigraphic approach, we have studied the evolutionary emergence of such genes at 19 phylogenetic levels. The majority of disease genes was already present in the eukaryotic ancestor, and the second largest number has arisen around the time of evolution of multicellularity. Conversely, genes specific to the mammalian lineage are highly underrepresented. Hence, genes involved in genetic diseases are not simply a random subset of all genes in the genome but are biased toward ancient genes.

FIG. 1.—

Phylogenetic framework used in the search for the human gene origins. Taxa represented in the databases with complete genomes or a substantial amount of TRACE and EST data are in bold. Taxa in italics are represented in the databases only with small numbers of highly conserved genes, and their exclusion from the analysis does not influence the results.

FIG. 2.—

Phylostratigraphy of all human genes and different classes of disease-causing genes. The total number of human genes (N = 22,845) found in the different phylostrata is plotted (blue line—squares, note logarithmic scale on the y axis). Distribution of the total number of evaluated disease genes (N = 1,760, red line—circles). The subsample of nonessential disease genes (N = 1,305), the stringent nonessential subsample (N = 1,020), and the genes involved in polygenic traits (N = 149) are also shown. The correlation coefficient between gene count and ranked evolutionary time is listed on top (estimated by Spearman's rank correlation coefficient).

FIG. 3.—

Probabilities of over- or underrepresentation of disease-causing genes in the respective phylostrata. Log-odds ratios show how the frequency of disease genes in each phylostratum deviates from the expected one estimated from the whole set of genes. Numbering of the phylostrata corresponds to those in figures 1 and 2 (*P < 0.05; **P < 0.01; ***P < 0.001, two-tailed hypergeometric test corrected for multiple comparison by FDR at 0.05 level).