SUMMARYSequence comparison allows the detailed analysis of evolution at the nucleotide and amino acid levels, but much less information is known about the structural evolution of genes, i.e. how the number, length and distribution of introns change over time. We constructed a parsimonious model for the evolutionary rate of intron loss (IL) and intron gain (IG) within the Brassicaceae and found that IL/IG has been highly dynamic, with substantial differences between and even within lineages. The divergence of the Brassicaceae lineages I and II marked a dramatic change in the IL rate, with the common ancestor of lineage I losing introns three times more rapidly than the common ancestor of lineage II. Our data also indicate a subsequent declining trend in the rate of IL, although in Arabidopsis thaliana introns continue to be lost at approximately the ancestral rate. Variations in the rate of IL/IG within lineage II have been even more remarkable. Brassica rapa appears to have lost introns approximately 15 times more rapidly than the common ancestor of B. rapa and Schenkiella parvula, and approximately 25 times more rapidly than its sister species Eutrema salsugineum. Microhomology was detected at the splice sites of several dynamic introns suggesting that the nonhomologous end-joining and double-strand break repair is a common pathway underlying IL/IG in these species. We also detected molecular signatures typical of mRNA-mediated IL, but only in B. rapa.

BackgroundThe majority of amino acid residues are encoded by more than one codon, and a bias in the usage of such synonymous codons has been repeatedly demonstrated. One assumption is that this phenomenon has evolved to improve the efficiency of translation by reducing the time required for the recruitment of isoacceptors. The most abundant tRNA species are preferred at sites on the protein which are key for its functionality, a behavior which has been termed “translational accuracy”. Although observed in many species, as yet no public domain software has been made available for its quantification.FindingsWe present here Seforta (Selection for Translational Accuracy), a program designed to quantify translational accuracy. It searches for synonymous codon usage bias in both conserved and non-conserved regions of coding sequences and computes a cumulative odds ratio and a Z-score. The specification of a set of preferred codons is desirable, but the program can also generate these. Finally, a randomization protocol calculates the probability that preferred codon combinations could have arisen by chance.ConclusionsSeforta is the first public domain program able to quantify translational accuracy. It comes with a simple graphical user interface and can be readily installed and adjusted to the user's requirements.