Prediksi Interaksi Drug Target pada Gen Kanker Menggunakan Metode Lasso-XGBoost

MABBI – Research conducted by Muh Fadhil Al-Haaq Gino, Wisnu Ananta Kusuma, and Mushthofa Mushthofa from IPB University entitled Prediksi Interaksi Drug Target pada Gen Kanker Menggunakan Metode Lasso-XGBoost.
Currently, cancer treatment is usually done with chemotherapy using chemical drugs that can cause side effects. An alternative treatment can use herbal compounds that known have fewer side effects. Drug Target Interaction analysis (DTI) can be performed to determine the interaction of herbal compounds with cancer proteins. In this study, a DTI prediction model is built by selecting features on the data set using Least Absolute Shrinkage and Selection Operator (LASSO) then data balancing performed with Synthetic Minority Oversampling Technique (SMOTE) and Extreme Gradient Boosting (XGBoost) performed to predict the interaction. The cancer-associated protein data were obtained from the Cancer Gene Census list, then the list used to search on the GDSC, DrugCentral and DrugBank databases to generate a list of drug compounds that interact with these proteins. In addition, plant compounds to be generated from the HerbalDB and Knapsack databases. Tests were performed on several types of feature extraction such as CTD, DC, PseAAC and PSSM. Predictive results suggest that several herbal compounds such as andrographolide, ursolic acid and oleanolic acid interact with cancer-associated proteins. In addition, LASSO-XGBoost was able to predict DTI in cancer with score of F1 0,861; AUROC 0,927; recall 0,857, precision 0,866; and accuracy 0,897. (Tri/MABBI)

Read more:


Leave a Reply

Your email address will not be published. Required fields are marked *