Hepatocellular carcinoma (HCC) is the most prevalent primary liver cancer, with a high mortality rate due to the limited therapeutic options. Systemic drug treatments improve the patient's life expectancy by only a few months. Furthermore, the development of novel small molecule chemotherapeutics is time-consuming and costly. Drug repurposing has been a successful strategy for identifying and utilizing new therapeutic options for diseases with limited treatment options. This study aims to identify candidate drug molecules for HCC treatment through repurposing existing compounds, leveraging the machine learning tool MDeePred. The Open Targets Platform, UniProt, ChEMBL, and Expasy databases were used to create a dataset for drug target interaction (DTI) predictions by MDeePred. Enrichment analyses of DTIs were conducted, leading to the selection of 6 out of 380 DTIs identified by MDeePred for further analyses. The physicochemical properties, lipophilicity, water solubility, drug-likeness, and medicinal chemistry properties of the candidate compounds and approved drugs for advanced stage HCC (lenvatinib, regorafenib, and sorafenib) were analyzed in detail. Drug candidates exhibited drug-like properties and demonstrated significant target docking properties. Our findings indicated the binding efficacy of the selected drug compounds to their designated targets associated with HCC. In conclusion, we identified small molecules that can be further exploited experimentally in HCC therapeutics. Our study also demonstrated the use of the MDeePred deep learning tool in in silico drug repurposing efforts for cancer therapeutics.
Keywords: MDeePred; drug candidate; drug repurposing; hepatocellular carcinoma; machine learning.