Improving Sequence Alignment using Class-Specific Score Matrices

Sequence alignment has been a most commonly used method in modern biological research. However, sequence alignment in "Twilight Zone" is still a challenging issue. Three protein class-specific matrices are constructed to investigate whether alignment in "Twilight Zone" can be improved by using class-specific matrices and gap penalties. The substitution pattern differences between different protein classes do improve the performances of the three class-specific matrices. Analyzing amino acids substitution pattern based on different matrices also suggests important information for protein sequence design.