Figure 1

Illustration of the method used to search for genes with evidence of exon shuffling Local alignments were performed between proteins A and B using a set of criteria to exclude the identification of paralogs (see Methods). Homologous regions were identified (aligned amino-acid regions are shown in blue). Four flanking windows were considered in nucleotide sequences for A and B, each one corresponding to one border of the protein alignment, and surrounding the projection of this border on the nucleotide sequences. Windows around the 5’ projected border extended from 20 nucleotides in the 5’ direction to 3 nucleotides in the 3’ direction, whereas windows around the 3’ projected border extended from 3 nucleotides in the 5’ direction to 20 nucleotides in the 3’ direction. The approximate reciprocal projections of these nucleotide windows on the protein sequences are highlighted in red. If the number of windows containing introns were at least three (for human and mouse) or at least two (for the remaining species), the genes coding for both proteins were considered as ES genes. If a protein had alignments to one or several non-homologous proteins but failed to meet the above mentioned minimum intron presence criterion in all these alignments, its gene was classified as an SS gene. If a protein had no alignments to non- homologous proteins, its gene was included in the WS category.