Disclosed herein is a driver gene signature for predicting survival in patients with solid tumors, such as hepatocellular carcinoma (HCC) and breast cancer. The gene signature includes ten tumor-associated genes, SH2D4A, CCDC25, ELP3, DLC1, PROSC, SORBS3, HNRPD, PAQR3, PHF17 and DCK. A decrease in DNA copy number or mRNA expression of SH2D4A, CCDC25, ELP3, DLC1, PROSC and SORBS3 in solid tumors is associated with a poor prognosis, while a decrease in DNA copy number or mRNA expression of HNRPD, PAQR3, PHF17 and DCK in solid tumors is associated with a good prognosis. Thus, provided herein is a method of predicting the prognosis of a patient diagnosed with HCC or breast cancer by detecting expression of one of more tumor-associated genes in a tumor sample and comparing expression of the one or more tumor-associated genes in the tumor sample to a control. Also provided is a method of treating a patient diagnosed with HCC or breast cancer by administering a therapeutically effective amount of an agent that alters expression or activity of one or more of the disclosed tumor-associated genes. Further provided are arrays comprising probes or antibodies specific for a plurality of tumor-associated genes or proteins.