A powerful way to identify driver genes with causal roles in carcinogenesis is to detect genomic regions that undergo frequent alterations in cancers. Here we identified 1,241 regions of somatic copy number alterations in 58 paired hepatocellular carcinoma (HCC) tumors and adjacent nontumor tissues using genome-wide single nucleotide polymorphism (SNP) 6.0 arrays. Subsequently, by integrating copy number profiles with gene expression signatures derived from the same HCC patients, we identified 362 differentially expressed genes within the aberrant regions. Among these, 20 candidate genes were chosen for further functional assessments. One novel tumor suppressor (tripartite motif-containing 35 [TRIM35]) and two putative oncogenes (hairy/enhancer-of-split related with YRPW motif 1 [HEY1] and small nuclear ribonucleoprotein polypeptide E [SNRPE]) were discovered by various in vitro and in vivo tumorigenicity experiments. Importantly, it was demonstrated that decreases of TRIM35 expression are a frequent event in HCC and the expression level of TRIM35 was negatively correlated with tumor size, histological grade, and serum alpha-fetoprotein concentration. Conclusion: These results showed that integration of genomic and transcriptional data offers powerful potential for identifying novel cancer genes in HCC pathogenesis. (HEPATOLOGY 2011;54:1227-1236) H epatocellular carcinoma (HCC) is the third leading cause of cancer-related mortality worldwide. New insights into the pathogenesis of this lethal disease are urgently needed. Chromosomal copy number alterations (CNAs) can lead to activation of oncogenes and inactivation of tumor suppressors in human cancers. 1 Thus, identification of cancer-specific CNAs will not only provide new insight into understanding the molecular basis of tumorigenesis but also facilitate the discovery of new cancer genes. 2,3 Using traditional methodologies, frequent DNA copy number gains at 1q, 8q, 7q, 17q, and 20q and losses at 4q, 8p, 13q, 16q, and 17p have been identified in HCC. 4,5 Although these technologies adequately detect some of the known candidate oncogenes or tumor suppressors, such as CHD1L (1q21) 6 and DLC1 (8p22), 7 their resolutions are limited for obtaining a comprehensive view of whole-genome copy number changes. High-density single-nucleotide polymorphism (SNP) arrays now provide the possibility of defining genome-wide copy number changes. 8,9 Additionally, there has been little progress in determining specific genes targeted by various common copy number gains and losses, in part due to limited availability of complementary transcriptional data on sufficient numbers of specimens to focus on a small list of candidate genes. Although several studies have been conducted to define potential cancer genes through combined analyses of genomic alterations and transcriptomes in HCC, they are constrained by the use of different sets and small sizes of tumor samples or by the use of relatively lower-resolution platforms. [10][11][12] In this study we applied a whole-genome SNP 6.0 array to define a comprehensive