Fig. 1

Prediction of new splicing sites and new exons, leading to the prediction of new transcripts. (a) Models of known transcripts. They concern the CDS part of the transcripts. They involve coding blocks, defining the nucleotidic sequences building a CDS, and functional sites delimiting exons on the gene. The coding blocks correspond to the intervals between the functional sites. A same block name occurring in several species indicates a conserved and orthologous region. For example, two transcripts known in mouse involve alternative exons denoted as C and BC, where both exons contain the block ‘C’, and block ‘B’ is an alternative 5’ extension of exon C. A known human CDS involves an exon BC estimated to be an ortholog of the mouse exon BC, and the known dog CDS involves an exon C, orthologous to the mouse exon C. (b) Gene model alignment. Each block and site in a gene is aligned with the gene sequence of an orthologous gene, resulting in pairwise gene alignments. These alignments reveal i) the orthology between already known sites (or coding blocks), ii) the sequence homology of known sites (or blocks) with not annotated loci in another gene, resulting in predicted sites and blocks (dashed bubbles). Here, aligning mouse and human genes revealed the presence in human of a homolog of the acceptor site of C. This site is thus declared as predicted. It indicates the human gene is able to express the C exon alone, without the ‘B’ part. Additional predictions: acceptor site of H in human and dog, and ‘B’ block plus its acceptor site in dog. The site graph summarizes pairwise orthology relations: a node is a functional site and an edge is an orthology relationship. (c) Predicted transcripts. Five transcripts are made possible from site and block predictions