Motivation: The analysis of tumor genomes is a key step to understand many aspects of cancer biology, ranging from its aetiology to its treatment or the oncogenic processes driving cell transformation. The first crucial step in the analysis of any tumor genome is the identification of somatic genetic variants that cancer cells have acquired during their evolution. For that purpose, a wide range of somatic variant callers have been created in recent years. However, it is still unclear which variant caller, or combination of them, is best suited to analyze tumor sequencing data. Results: Here we present a study to elucidate if different variant callers (MuSE, MuTect2, SomaticSniper, VarScan2) and strategies to combine them (Consensus and Union) lead to different downstream results in these three important aspects of cancer genomics: driver genes, mutational signatures and clinically actionable targets identification. To this end, we assess their performance in five different projects from The Cancer Genome Atlas (TCGA). Our results show that variant calling decisions have a significant impact on these downstream analyses, rendering important differences in driver genes prediction and mutational status among variant call sets, as well as in the identification of clinically actionable targets. More importantly, it seems that there is not a one-size-fits-all variant calling strategy, as the optimal decision seems to depend on both, the cancer type and the goal of the analysis.