Abstract
Drug-induced liver injury (DILI) is the major cause of clinical trial failure and postmarketing withdrawals of approved drugs. It is very expensive and time-consuming to evaluate hepatotoxicity using animal or cell-based experiments in the early stage of drug development. In this study, an in silico model based on the joint decision-making strategy was developed for DILI assessment using a relatively large dataset of 2608 compounds. Five consensus models were developed with PaDEL descriptors and PubChem, Substructure, Estate, and Klekota–Roth fingerprints, respectively. Submodels for each consensus model were obtained through joint optimization. The parameters and features of each submodel were optimized jointly based on the hybrid quantum particle swarm optimization (HQPSO) algorithm. The application domain (AD) based on the frequency-weighted and distance (FWD)-based method and Tanimoto similarity index showed the wide AD of the qualified consensus models. A joint decision-making model was integrated by the qualified consensus models, and the overwhelming majority principle was used to improve the performance of consensus models. The application scope narrowing caused by the overwhelming majority principle was successfully solved by joint decision-making. The proposed model successfully predicted 99.2% of the compounds in the test set, with an accuracy of 80.0%, a sensitivity of 83.9, and a specificity of 73.3%. For an external validation set containing 390 compounds collected from DILIrank, 98.2% of the compounds were successfully predicted with an accuracy of 79.9%, a sensitivity of 97.1%, and a specificity of 66.0%. Furthermore, 25 privileged substructures responsible for DILI were identified from Substructure, PubChem, and Klekota–Roth fingerprints. These privileged substructures can be regarded as structural alerts in hepatotoxicity evaluation. Compared with the main published studies, our method exhibits certain advantage in data size, transparency, and standardization of the modeling process and accuracy and credibility of prediction results. It is a promising tool for virtual screening in the early stage of drug development.
Highlights
Bringing a new drug to market is a time-consuming and expensive process
To further explore the chemical diversity of the dataset, the Tanimoto similarity index of the modeling dataset was calculated using PubchemFP, Substructure fingerprint (SubFP), and KRFP, respectively. e heat map of the Tanimoto similarity index based on PubchemFP is shown in Figure 13. e average Tanimoto similarity index calculated from PubchemFP, SubFP, and KRFP is 0.335, 0.372, and 0.165, respectively
SVM: support vector machine; GA-SVM: genetic algorithm-support vector machine; RF: random forest; adetailed information of these methods can be found in [40]; NB: naive Bayes; kNN: k-nearest neighbor; CT: classification tree; Nm is the number of compounds in the modeling dataset; Nde is the number of descriptors/fingerprints used in (Q)SAR models; CV: cross-validation; CV5: fivefold cross-validation; EV: external validation; TV: validation on the test set; CSM: consensus model; JDM: joint decision-making model
Summary
Bringing a new drug to market is a time-consuming and expensive process. Lots of candidate drugs fail to become drugs or withdraw from market mainly because of their safety and lack of efficacy [1]. Taking into consideration the above-mentioned issues, we attempt to develop a joint decision-making model based on a relatively large and chemically diverse DILI dataset to achieve better performance and credibility, as well as a wide AD. To achieve this goal, a total of 2608 compounds (1643 DILI positives/965 DILI negatives) were first manually collected. We hope the model could be a useful tool for DILI assessment of drug candidates in the early stage of drug discovery, and the privileged substructures will be helpful for pharmaceutical chemists to design an appropriate structure for drug molecules
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
