A Novel Feature Vector for Short Coding Region Recognition of Human Gene

Protein coding region recognition has been the one of the main topics in computational biology. In this study, we try to solve the issue using new methods. We received two new features by integrating the information of the distributions of stop codons and the information of base compositional bias. And the pseudo-base composition features were given, which can extract the information of the bases interaction in different positions. The accuracy of the algorithm was tested based on a large database of human genes. The average accuracy achieved by three new features was as high as 92.73% for the fragments with length of 192 base pairs. And the accuracy of the algorithm with 15 features can achieve 95.65% in the same length. We find that the use of the combination of two characters and pseudo-base composition features improve the accuracy of coding region recognition.