A scoring function development for protein structure classification based on sequence and structure information

This project is to combine sequence and structure information of proteins to set up scoring function to classify protein structures whose belongings are unknown. The scoring function is basically explained as distance between two proteins. We began this project with searching useful informations and construct geometric and topological representations and distance metrics of those useful informations. In this paper, we used sequence information, residual number information,and Amino Acid information. Then through learning from 40 classified proteins and linear programming, we try to determine the weights for each piece of informations. Ideally, those ”good” pieces of information are supposed to win higher weights and less good ones have lower weights. The scoring function with weights determined can be used to predict classification of other proteins whose structures are unknown.