Summary: | 碩士 === 國立臺灣大學 === 醫學工程學研究所 === 91 === Now there are lots of protein classification databases to do protein classification by sequence similarity, structure identity, or function similarity, like CATH, SCOP, and FSSP. In previously literature, it is pointed out that it was necessary to take protein structure evolution into consideration when doing protein structure classification. There are four main parts in protein structure evolutions :
Insertion/deletion/substitution
Circular permutation
Strand Invasion/Withdrawal
Beta-Hairpin Flip/Swap
However the circular permutation is one ideal of the protein structure evolution that can be solved by computer algorithm.
Circular permutation allows two proteins with different sequence and structure to become similar or the same by rearrangement. If two proteins have the appearance of such thing, we call that these two proteins have circular permutation and it also plays an important role in protein classification. In 1997, one literature found two different proteins, Saposin and Swaposin, that have circular permutation for each other and they came from the same ancestry. By the definition of protein evolution, the function of these two proteins should be the same. Which is to say that these two proteins should be clustered in the same fold.
In this paper, we added the ideal of circular permutation and took CATH to be a consultation and used Genetic Algorithm to do re-classification. By the definition of fitness function and quality in Genetic Algorithm, our research is better than CATH.
|