The mRNA vaccines are considered to be effective treatment strategies for cancers, but its progress in chronic hepatitis B virus (HBV) related-hepatocellular carcinoma (HCC) was slow. This study aimed to find potential antigens and identify suitable patients in HBV related-HCC for guiding mRNA vaccine development. We integrated the transcriptome RNA expression matrices and somatic mutation data from TCGA and ICGC datasets. A consistency matrix was constructed by using ConsensusClusterPlus to identify the immune subtypes. Graph learning based dimensional reduction was analyzed to establish immune landscape. Four upregulated and mutated antigens (EPS8L3, TCOF1, EZH2, and NOP56) were highly correlated with unfavorable clinical outcomes and antigen presenting cells (APCs). And two distinct immune phenotypes with differential clinical, cellular, and molecular characteristics were identified by in the ICGC and TCGA cohorts. IS1 is immune “hot” and immunosuppressive phenotype, with low tumor mutation burden (TMB) and high immune checkpoints (ICPs). On the contrary, IS2 is immune “cold” phenotype with high TMB and low ICPs. Monocle3 package was used to further study the intra-cluster heterogeneity, which identified cluster IS2A/2B within IS2 subtype was determined to be more suitable for mRNA vaccine. In summary, EPS8L3, TCOF1, EZH2, and NOP56 are potential antigens for mRNA vaccine development against HBV related-HCC, and patients in IS2A/2B are relatively more suitable for vaccination.