Prediction of Thermophilic Proteins Using Voting Algorithm

Thermophilic proteins have widely used in food, medicine, tanning, and oil drilling. By analyzing the protein sequence, the superior structure and properties of the protein sequence are obtained, which is used to efficiently predict the protein species. In this paper, a voting algorithm was designed independently. Protein features and dimensions were extracted and reduced, respectively. Data was predicted by WEKA. Next, the voting algorithm was applied to the data obtained by the above processing. In this experiment, the highest accuracy rate of 93.03% was achieved. This experiment has at least two advantages: First, the voting algorithm was developed independently. Second, any optimization method was not used for this experiment, which prevents over-fitting. Therefore, voting is a very effective strategy for the thermal stability of proteins. The prediction data set used in this paper can be freely downloaded from http://lab.malab.cn/~lijing/thermo_data.html.