Methods for evaluating exon-protein correspondences

• Published in: Bioinformatics 1995-10, Vol.11 (5), p.509-515
• Main Authors: Stoltzfus, Arlin, Spencer, David F., Doolittle, W.Ford
• Format: Article
• Language: English
• Subjects: Algorithms; Biological and medical sciences; Biological Evolution; Databases, Factual; Exons; Fundamental and applied biological sciences. Psychology; General aspects; Genetic Techniques; Introns; Mathematics in biology. Statistical analysis. Models. Metrology. Data processing in biology (general aspects); Models, Genetic; Molecular Structure; Proteins - chemistry; Proteins - genetics; Software; Triose-Phosphate Isomerase - chemistry; Triose-Phosphate Isomerase - genetics
• ISSN: 1367-4803; 0266-7061; 1460-2059
• DOI: 10.1093/bioinformatics/11.5.509

According to the exon theory of genes, protein-coding genes evolved originally by combinatorial assembly of mini-gene precursors of modern exons. If so, then exons should tend to encode discrete bits of protein structure, as first suggested by C.C.F.Blake. In order to assess the evidence for Blake's conjecture, we have developed methods for evaluating the significance of correspondences between split gene structure and protein structure, using computer programs for measuring observed correspondences and comparing them to random expectations. Initial results of applying these methods to data on ancient proteins have been presented elsewhere. Here we describe the algorithms in detail, and demonstrate their effectiveness in finding correlations in idealized test cases. The likely effects of deletion and putative displacement (‘sliding’) of introns on the ability to detect correlations are also examined.