Breast cancer drug toxicity prediction Based on AdaBoost Extremely Random Tree

Abstract

Estrogen Receptor α (ERα) is considered as an important target for treating breast cancer, so compounds that can antagonize ERα may be candidate drugs for breast cancer. We predict the toxicities of candidate compounds by machine learning to achieve the virtual screening of breast cancer drugs. In order to improve the performance of the evaluation for toxicities of drugs in virtual screening, a toxicity prediction method that integrates an adaptive boosting extremely random tree algorithm is proposed. We analyze the function of adaptive factors in the algorithm and apply the improved algorithm to predict the toxicity of breast cancer drugs. The experimental results show that the proposed method can accurately predict the toxicities of breast cancer drugs, and increase the efficiency of drug discovery in the early stage based on virtual screening.