Cancer immunotherapy, including immune checkpoint inhibitors (ICIs) and chimeric antigen receptor T-cell (CAR-T) therapy, has revolutionized the paradigm in cancer treatment. However, the clinical outcome of immunotherapy varies considerably among patients and only a minority of patients achieve long-term clinical benefits. This is largely attributed to the fact that existing cancer immunotherapies, which concentrate on several classical targets (CTAL-4, PD-1/PD-L1, etc.) and limited types of immune cell populations (T cells), are insufficient to cope with the complexity of highly heterogeneous tumor microenvironment (TME). This calls for more efforts to not only expand our toolbox for manipulating anticancer immunity but also diversify our combinational strategies. To this end, it is urgent to deeper our understanding of cancer immunotherapy by using both experimental and computational methodologies from multi-scale perspectives: 1) novel targets from either tumor cells or non-tumor cells within TME (e.g., tumor intrinsic resistance drivers, new immune checkpoints, neoantigens), 2) in-depth characterization of more immune cell populations (e.g., macrophages, Tregs, B cells) and their interactions and dynamics within TME, 3) landscape of actionable targets in patient populations for combination design. These efforts will open the avenue of rational design of combinational immunotherapies, allowing researchers and clinicians to design novel targeting therapeutics or to optimally orchestrate combinatory strategies aiming to surmount resistance mechanisms and improve clinical outcomes.