Identifying Candidates for Breast Cancer Using Interactions of Chemicals and Proteins

Breast cancer is one of the major cause of cancer death in women worldwide. Therefore, it is urgent to discovery novel drugs or design effective treatments for this disease. However, the research and development of drugs by using only experimental methods is always time-consuming and expensive. With the development of computer science, some advanced computational methods can make full use of known knowledge to design candidate drugs, thereby reducing the cost and time of experimental testing. In this study, a computational method was proposed to identify novel candidates for breast cancer. The approved drugs and genes of breast cancer were taken as the input of the method. The chemical-chemical interactions and chemical-protein interactions were adopted to extract possible candidates from large numbers of existing chemicals. The method included three stages, termed searching stage, filtering stage and selecting stage. In the searching stage, chemicals that have associations with approved drugs were extracted. Then, these chemicals were screened in the filtering stage to discard those that have no relationships with breast cancer related genes. Finally, a clustering algorithm, termed as EM clustering algorithm, was employed to identify the potential candidates in the selecting stage. An extensive analysis by retrieving literature indicated that multiple selected candidates, such as gefitinib, canertinib and sirolimus, that have been approved for other diseases were confirmed to have anti-breast cancer activities. Therefore, this method can provide some valuable instructions for drug repositioning.