Gene mutations are closely related to cancers and drug sensitivity. Noninvasive liquid biopsy was used to detect mutations of ctDNA in plasma, which is regarded as an indicator of chemotherapy reaction. In this study, we performed exon sequencing of 416 cancer-related genes for cancer primary tissue and plasma samples of 20 patients in 11 cancers. The comprehensive mutation landscape was obtained by bioinformatics tools. In all samples, a total of 0–135 genes involved somatic mutations, and 5–209 genes involved copy number variation. APC, KRAS, and TP53 were detected as frequently mutated genes. Nineteen genes with high-frequency copy-number amplification and 59 with frequent copy-number deletions were identified. By quantitatively assessing the degree of agreement, we found that liquid biopsy is reliable instead of tissues. Besides, 31 mutation prognostic markers in 7 cancers were screened by integrating the consistent mutations and enlarging samples in TCGA. Moreover, from drug-mutation network, 25 drugs connected with 9 mutations (B-Mut-9) were obtained which can be served as drug biomarkers in blood. This was proved by further integrating the mutation information of patients in TCGA into drug-mutation network. In summary, the variation in ctDNA can be used as the biomarkers for cancer prognosis and drug efficacy prediction. Impact statement Gene mutations are closely related to cancers and drug sensitivity and noninvasive liquid biopsy was used to detect mutations of ctDNA in plasma. In this study, we performed exon sequencing of 416 cancer-related genes for cancer primary tissue and plasma samples of 20 patients in 11 cancers and obtained the comprehensive mutation landscape. We found that liquid biopsy is reliable in place of tissue biopsy. And 31 potential unique mutation prognostic markers were screened in 7 cancer types. Moreover, the drug-mutation network (DMN) was constructed and 9 gene mutations (B-Mut-9) were confirmed that can be served as drug biomarkers in blood. Our study showed that the variation in ctDNA can be used as the biomarkers for cancer prognosis and drug efficacy prediction. This can provide a reference for clinical noninvasive testing.