The most important question in cancer genomics research is a biological understanding of large data sets generated by high-throughput technologies such as whole genome sequencing and RNA-Seq. As a result, a typical omics-based cancer discovery begins with methylation analysis (epigenetic level), followed by differential gene expression (mRNA level), miRNA/ncRNA expression profiling (non-coding transcript level), differential protein expression and phosphorylation analysis (proteomics level), and finally metabolomics analysis (metabolite level). These analyses typically yield a list of genes that are differentially expressed/methylated in multiple groups. Furthermore, genome-wide association studies, genome/exome sequencing, and genes with loss/gain-of-function variants are widely used to track the activity of thousands of genes, generating gene lists for further analysis. Therefore, gaining biological insight from the generated gene list is the primary challenge to realising the potential of these technologies.

Tumor suppressor genes (TSGs) and oncogenes are two types of genes that play opposing roles in cancer development. In brief, oncogenes are specific genes that can transform cancer cells, stimulate cell proliferation, and promote cell survival by interfering with apoptosis. TSGs inhibit cell proliferation and tumor development by acting on the inverse side of cell growth control. In many tumors, these genes are lost or inactivated, removing negative regulators of cell proliferation and contributing to tumour cell abnormal proliferation. In addition to TSGs and oncogenes, an increasing number of driver genes have been discovered to play both tumor suppressor and oncogene roles depending on the biological context, such as epigenetic regulation, interacting partners, cancer type, and tissue specificity. A comprehensive investigation of TSGs and oncogenes may expand our understanding of the relationship between TSGs and OCGs in cancer development. Our TSGene 2.0 database firstly complied a list of 73 TSGs that also play oncogenic roles in various cancers. As the integrative resource for tumor suppressors and oncogenes, POTSON offers the most comprehensive gene list (251 known genes) with dual role based on literature information. As a result, a typical POTSON analysis began with the generation and integration of omics data. After generating the gene list outside of POTSON, the user could use our POTSON to map the gene and generate an intelligent summary. The following are the key features.

Please check the Data menu.

Please check the Intelligent summary menu.

Please check the Dashboard menu.

The human TSGs and oncogenes were retrieved from five different databases, including: TSGene, ONGene, OncoKB, Cancer Gene Census, and Network of Cancer Genes. TSGene and ONGene, in general, are specific for TSGs and oncogenes with literature information. TSGs and oncogenes have also been collected from the OncoKB, Cancer Gene Census, and Network of Cancer Genes databases. However, some OncoKB genes are missing literature information. We made every effort to connect the relevant literature to the genes. Based on 18,295 PubMed abstracts, a total of 2285 genes in POTSON were identified and mapped to the most recent records in the NCBI Gene database. The basic gene information, including the official gene symbol, gene type, alias, chromosome and map locations, was obtained from the NCBI Gene database.

The 251 dual role genes were defined based on the known TSGs and oncogenes from the five databases. The associated literature and cancer-type information were assigned automatically as well. It should be noted that:

To date, three tools have been developed to predict tumour suppressive and oncogenic roles: TUSON (Tumor Suppressor and OG Explorer), DORGE (Discovery of Oncogenes and tumor suppressoR genes using Genetic and Epigenetic features), and 20/20+ machine-learning method. Specifically, TUSON and 20/20+ can predict protein-coding TSGs and ONGs using mutational signatures from large-scale cancer genomics datasets. The DORGE employed both genetic and epigenetic characteristics. In addition to these three tools, we integrated the following six tools for cancer driver gene prediction as well.

• MutsigCV is a model that use mutational covariates to identify genes that have been significantly mutated in cancer genomes.
• MuSiC stands for Mutational Significance in Cancer.
• OncodriveCLUST is designed to identify genes whose mutations favour large spatial clustering.
• OncodriveFML is a general framwork for identifying cancer driver mutations in coding and non-coding regions.
• OncodriveFM is the predecessor to OncodriveFML.
• ActiveDriver is a novel method for identifying genes that have significant phosphorylation-related mutations.