Machine Learning Screens Potential Drugs Targeting a Prognostic Gene Signature Associated With Proliferation in Hepatocellular Carcinoma

• DOI: 10.3389/fgene.2022.900380
• Publication Date: 2022-06-28T04:45:28Z
• Authors: Jun Liu; Jianjun Lu; Wenli Li; Wenjie Mao; Yamin Lu
• Keywords: Gene signature; Univariate

Background: This study aimed to screen potential drugs targeting a new prognostic gene signature associated with proliferation in hepatocellular carcinoma (HCC). Methods: CRISPR Library and TCGA datasets were used explore differentially expressed genes (DEGs) related the of HCC cells. Differential expression analysis, univariate COX regression random forest algorithm multiple combinatorial screening construct signature. Then predictive power was validated ICGC datasets. Furthermore, this screened. Results: A total 640 DEGs identified. Using Cox analysis algorithm, 10 hub Subsequently, using multiplex screening, five (FARSB, NOP58, CCT4, DHX37 YARS) Taking median risk score as cutoff value, patients divided into high- low-risk groups. Kaplan-Meier performed training set showed that overall survival high-risk group worse than (p < 0.001). The ROC curve good efficiency (AUC > 0.699). mutation, cancer cell stemness immune function changes. Prediction immunotherapy suggetsted IC50s checkpoint inhibitors including A-443654, ABT-888, AG-014699, ATRA, AUY-922, AZ-628 lower those group, while AMG-706, A-770041, AICAR, AKT inhibitor VIII, Axitinib, AZD-0530 higher group. Drug sensitivity indicated FARSB positively correlated Hydroxyurea, Vorinostat, Nelarabine, Lomustine, negatively JNJ-42756493. Raltitrexed, Cytarabine, Cisplatin, Tiotepa, Triethylene Melamine. YARS Fluphenazine Megestrol acetate. NOP58 Vorinostat 6-thioguanine. CCT4 Nerabine. Conclusion: five-gene can be for prediction stratification patients. Potential deserve further attention treatment HCC.