Prediction of Drug Targets Related to HCC Metastasis from the Perspective of Programmed Cell Death Based on Transformer

Hepatocellular carcinoma (HCC) ranks as the sixth most prevalent cancer globally and the third leading cause of cancer-related deaths. The early diagnosis of HCC is challenging, and its propensity for metastasis results in generally poor prognosis. Existing studies have demonstrated a close association between HCC metastasis and programmed cell death (PCD). However, due to the high-dimensional, high-order nonlinear interactions and complex regulatory mechanisms of genes, establishing the biological process from genes to PCD to HCC metastasis through biological experiments is difficult. These genes represent crucial drug targets for blocking HCC progression. Deep learning technologies, particularly Transformer models, have shown great potential in analyzing clinical and genetic data in oncology. In this study, we identified 147 key genes related to PCD that are closely associated with HCC metastasis and developed a Transformer-based predictive model for HCC metastasis. To elucidate the potential of these genes as drug targets, we first investigated the impact of various types of PCD on the pathophysiology of HCC. Subsequently, we validated the drug responses of different patients through immunoassays and drug sensitivity tests. Finally, survival analysis indicated that these genes significantly affect patient survival rates. In summary, we have identified numerous drug targets influencing HCC metastasis through PCD.