Structure of vertebrate genes: a statistical analysis implicating selection.

This paper conducts a statistical analysis of the size distribution of exons and six other gene parts [the transcription unit, introns, intervening DNA (sum of introns), mRNA (sum of exons), and leader and trailer regions of mRNA] as well as the number of exons, the percentage of introns, the placement of introns within the gene, and the potential for frameshifts from coding exon shifts. The first seven variables measured in base pairs fit log-normal distributions. Significant correlations between the sizes of intervening DNA and mRNA, the sizes of leader and trailer regions, and the sizes of introns and flanking exons exist. Introns occur at nonrandom frequencies within the codon frame, in untranslated regions, and relative to the frameshift potential from exon movement or duplication. These nonrandom patterns in gene structure demonstrate that models of gene evolution must incorporate selective processes.