Email updates

Keep up to date with the latest news and content from BMC Bioinformatics and BioMed Central.

Open Access Methodology article

The effects of multiple features of alternatively spliced exons on the KA/KS ratio test

Feng-Chi Chen12 and Trees-Juen Chuang1*

Author Affiliations

1 Genomics Research Center, Academia Sinica, Academia Road, Nankang, Taipei 11529, Taiwan

2 Division of Biostatistics and Bioinformatics, National Health Research Institute, Miaoli County 350, Taiwan

For all author emails, please log on.

BMC Bioinformatics 2006, 7:259  doi:10.1186/1471-2105-7-259

Published: 19 May 2006



The evolution of alternatively spliced exons (ASEs) is of primary interest because these exons are suggested to be a major source of functional diversity of proteins. Many exon features have been suggested to affect the evolution of ASEs. However, previous studies have relied on the KA/KS ratio test without taking into consideration information sufficiency (i.e., exon length > 75 bp, cross-species divergence > 5%) of the studied exons, leading to potentially biased interpretations. Furthermore, which exon feature dominates the results of the KA/KS ratio test and whether multiple exon features have additive effects have remained unexplored.


In this study, we collect two different datasets for analysis – the ASE dataset (which includes lineage-specific ASEs and conserved ASEs) and the ACE dataset (which includes only conserved ASEs). We first show that information sufficiency can significantly affect the interpretation of relationship between exons features and the KA/KS ratio test results. After discarding exons with insufficient information, we use a Boolean method to analyze the relationship between test results and four exon features (namely length, protein domain overlapping, inclusion level, and exonic splicing enhancer (ESE) frequency) for the ASE dataset. We demonstrate that length and protein domain overlapping are dominant factors, and they have similar impacts on test results of ASEs. In addition, despite the weak impacts of inclusion level and ESE motif frequency when considered individually, combination of these two factors still have minor additive effects on test results. However, the ACE dataset shows a slightly different result in that inclusion level has a marginally significant effect on test results. Lineage-specific ASEs may have contributed to the difference. Overall, in both ASEs and ACEs, protein domain overlapping is the most dominant exon feature while ESE frequency is the weakest one in affecting test results.


The proposed method can easily find additive effects of individual or multiple factors on the KA/KS ratio test results of exons. Therefore, the system can analyze complex conditions in evolution where multiple features are involved. More factors can also be added into the system to extend the scope of evolutionary analysis of exons. In addition, our method may be useful when orthologous exons can not be found for the KA/KS ratio test.