Abstract. Osteosarcoma, which is the most common type of highly malignant bone tumor in children and adolescents, has poor diagnosis and 2-year survival rates of 15-20% following surgery or radiotherapy, and has therefore generated marked attention. In order to investigate the potential biomarkers for diagnosing osteosarcoma, the expression profiling data from normal and disease tissues were compared, respectively, and the differentially-expressed genes were analyzed by three different statistical tests. Interacting proteins were determined and an interaction network was constructed by Search Tool for the Retrieval of Interacting Genes database. Subsequently, the protein interaction network was decomposed and Gene Otology annotation using Cytoscape, Mcode and Bingo, was conducted on the function modules. Finally, three differentially-expressed genes GJA1, COL1A2 and COL5A2 were identified, and an interaction network was successfully generated with COL1A2 and COL5A2 at the core. From the results, it was observed that COL1A2 and COL5A2 interact with a number of genes of the matrix metalloprotease (MMP) family, including MMP1, MMP2, MMP3 and MMP14, TGFβ and RUNX2. Furthermore, these genes have been confirmed to be important in the tumorigenesis of osteosarcoma. It was hypothesized that the upregulation of the COL gene family may be considered as a diagnostic marker for osteosarcoma and collagen may be administered as a therapy.