A Novel Ion Channel Prediction Method Based on XGBoost Model

摘要

Nowadays, ion channels are associated with many diseases, including cancer. Ion channels are pore-forming proteins in many creatures. They can help to fight infections, encode memory and learning and even enable pain signals. Based on a dataset of Ion channels released by University of Liverpool, the relationship between electric signals and ion switching can be explored. In this paper, a novel ion channel prediction method based on XGBoost model was proposed. Besides, data cleaning and feature engineering were also included. Finally, several classical methods are compared with the proposed XGBoost model in the results section. The results show the superiority of the proposed XGBoost model, which can reach 0.943 in accuracy and 0.982 in micro F1 score. The method may have a far-reaching impact on many areas related to cell health and migration.