Specialized web-based tools are available to enable investigators to query, retrieve and analyze cancer-related data in a rapid, reliable and efficient manner. The Cancer Genome Anatomy Project (CGAP) (14) of NCI includes a number of bioinformatic analysis tools and interconnected modules that enable users to access CGAP data. These data include cancer-relevant genes and SNPs, malignant tissues and chromosomal aberrations in cancer patients. Moreover, CGAP provides information regarding the differential expression of a given gene in normal, precancerous and cancerous tissues based on Serial Analysis of Gene Expression (SAGE), as well as RNA interference (RNAi) constructs that target cancer-related genes, and diagrams of biochemical pathways and protein complexes. The UCSC Cancer Genomics Browser (15,16) is a suite of web-based tools used to integrate, display and analyze cancer genomics and clinical data. The browser allows whole-genome views of several different types of genomics and associated clinical data. Various datasets can be viewed together as coordinated ‘heatmap tracks’, thus enabling the user to make comparisons across studies and cancer types. Annotated biological pathways, collections of genes, genomic or clinical information can be sorted, filtered, aggregated, classified and viewed interactively based on any given feature set, including clinical features, annotated biological pathways and user-contributed collections of genes. The Cancer Genome Workbench (CGWB) (17) includes copy number, mutation, expression and methylation data from various projects, including TCGA, the Catalogue of Somatic Mutations in Cancer (COSMIC) (18,19), Johns Hopkins University, and the Therapeutically Applicable Research To Generate Effective Treatments (TARGET) initiative. CGWB provides a series of tools for visualizing genomic and transcription alterations from different cancer samples. The data in CGWB can be viewed in three different ways: i) Integrated track, which provides a sample-level view of genomic alterations from multiple data sources; ii) Heatmap view, an interactive graphical view of gene expression and copy number data and their associated clinical features; and iii) Bambino, an alignment viewer for NGS data.

Several repositories of driver genes or gene families that play a causal role in carcinogenesis have been developed. The Tumor Gene Family Databases (TGDBs) (20) contain a broad range of information regarding genes involved in cancer. Apart from TGDB itself, the data of two component databases, the Oral Cancer Gene Database (OrCGDB) (21) and the Breast Cancer Gene Database (BCGD) (22), have been merged into TGDBs. Gene information includes gene aliases, cell location, biochemical function, frequency in various tumors, oncogenicity, chromosomal location, tumor gene type (either proto-oncogene or tumor suppressor gene) and the signal transduction pathways in which the gene of interest is involved.

The DriverDB database (23) compiles a large amount (>6,000 cases) of exome-sequencing (exome-seq) data, annotation databases such as dbSNP (24), 1000 Genome (25) and COSMIC, as well as various bioinformatics algorithms for the identification of driver genes or mutations. The database can be queried either by cancer type, where the driver genes/mutations for a specific cancer type are estimated, or by gene where the mutation information of a driver gene in five different aspects is presented. Meta-Analysis, another option offered in DriverDB, enables users to identify driver genes in custom-defined samples according to clinical criteria. The RAS Oncogene Database (RASOnD) (20) integrates large amounts of genomics and proteomics data derived from publicly available databases such as NCBI’s GenBank (26), Online Mendelian Inheritance in Man (OMIM) (27), Universal Protein Resource (UniProt) (28), Protein Databank (PDB) (29), Kyoto Encyclopedia of Genes and Genomes (KEGG) (30) and PubMed (31). The RASOnD database contains 199,046 entries from 101 species, allowing investigators to retrieve information regarding RAS oncogene SNPs, chromosomal positions, disease associations and nucleotide and amino acid positions.

Cancer is characterized by abundant genetic abnormalities in the form of mutations, SNPs, copy number alterations (CNAs), genomic rearrangements and gene fusions. To manage the increasing amount of information, public resources have been implemented to collect, curate, annotate and analyze data regarding cancer genetic variations. COSMIC (18,19) is the largest public database that stores and displays information on somatically acquired mutations involved in cancer and associated clinical and phenotypic data. Currently, COSMIC contains information on 28,735 genes, 2,002,811 coding mutations and 10,435 fusion gene mutations reported in 1,029,547 cancer samples. The data are primarily extracted from published scientific literature and whole-genome sequencing screens from CGP. To provide a uniform representation of the data, a histology and tissue ontology has been created. COSMIC uses the BioMart (32) data mining software that enables users to filter the available data according to cancer sample, gene, mutation, tumor site, histology and tumor. The results are presented in tabulated format (Fig. 2).

The Cancer Gene Census (CGC) (33) lists >1% of all human genes which bear mutations that causally contribute to carcinogenesis. The Gene Census data include gene symbol according to HGNC (34), a short description of the gene, gene chromosomal location, type of mutations (i.e., somatic or germline), type of tumor and the cancer syndrome in which the mutated gene is involved. The data are provided in a table and can be downloaded and exported in several formats. The catalogue is updated at regular intervals. BioMuta (35) is a curated database of cancer-related non-synonymous single-nucleotide variations (nsSNVs) that affect functional sites. The datasets are derived from the TCGA, COSMIC, ClinVar (36) and UniProt Knowledgease (UniProtKB) (37) databases. Due to the large amount of data present in the primary NGS repositories, the High-performance Integrated Virtual Environment (HIVE) platform (35) has been implemented in BioMuta in order to store, analyze, compute and curate NGS data and associated metadata. CaSNP (38) is a comprehensive collection of CNAs from 11,485 Affymetrix SNP arrays with raw data from NCBI’s Gene Expression Omnibus (GEO) (39), additional arrays from the TCGA consortium and a few individual publications covering 34 different cancer types in 105 studies. The user can query CaSNP by gene, region or cancer type and retrieve information regarding the frequencies of copy number aberrations for each study. CaSNP also provides a heatmap showing CNAs estimated at each SNP marker around the query region across all studies. CanProVar (40) has been developed to store and prevent germline and somatic amino acid variations in the human proteome associated with human tumorigenesis based on published literature. The CanGEM (41) database stores clinical information on tumor samples and array Comparative Genomic Hybridization (aCGH) data to detect gene CNAs in cancer. Users can create custom datasets for specific clinical sample characteristics or CNAs of individual genes. The Integrative Cancer Profiler System (ICPS) (10) database integrates genomic alterations such asCNA and LOH, with transcription signatures (SAGE, microarray) in order to study gene profiles in one or more different types/subtypes of cancer. Currently, ICPS contains five different data types and 23,375 experiments covering 11 major cancer types. Apart from public data, ICPS also supports in-house data of users.

Given that ~50% of human tumors harbor TP53 gene mutations (42), a UMD TP53 (43) database was created to provide detailed information on TP53 mutants such as the molecular and cell properties of each TP53 mutant and localization or various gains of functions. The UMD TP53 database contains >110,000 entries. The International Agency for Research on Cancer (IARC) TP53 (44,45) is a comprehensive resource that compiles all TP53 gene variations in human cancers derived from scientific publications. The datasets available in the resource are: TP53 somatic and germline mutations, validated common TP53 polymorphisms identified in human population and their functional and clinical impact, TP53 gene status (i.e., wild-type, mutant, null) in various human cell lines, mouse models with engineered TP53 constructs and experimentally induced TP53 mutations.

Epigenetic modifications, such as DNA methylation and chromatin-modifying factors, play a critical role in carcinogenesis by regulating tumor-suppressor gene silencing, proto-oncogene activation and chromosomal instability (46). MethyCancer (47), the database of human DNA methylation and cancer, was developed to study the association of DNA methylation, gene expression and cancer. It contains data of DNA methylation, cancer-relevant genes and CpG Island (CGI) clones derived from high-throughput sequencing. The MethyView option allows the graphical presentation of CGI information of >30,000 genes. PubMeth (48) is a cancer methylation database that includes information of genes that have been reported in the literature to be methylated in various cancer types. The information is extracted from PubMed abstracts using a text-mining approach, GoldMine, followed by manual annotation. There are two options for searching the database, the ‘gene-centric’ (the cancer types/subtypes where the genes of the interest are reported to be methylated) and the ‘cancer-centric’ (the genes reported to be methylated in a particular cancer type/subtype). In ChromoHub V2 (49), chemical, structural and biological data extracted from public repositories, such as TCGA and ICGC, are mapped on phylogenetic trees of protein families involved in chromatin-mediated signaling.

OncomiRs, microRNAs that are associated with diverse cancer-related processes, play a significant role in the epigenetic regulation of cancer. The miRCancer database (50) provides a comprehensive collection of microRNA expression profiles in various human malignancies that are automatically extracted from publications in PubMed. OncomiRDB (51) is a database developed for annotating the experimentally validated oncomiRs from literature. The database includes 2,259 entries of oncomiR regulations, covering 328 miRNAs and 829 target genes in 25 cancer tissues extracted from published literature. The user is able to search by miRNA, tissue, tumor, target gene and function (e.g., proliferation, apoptosis, migration).

Databases have been designated to extract, store and interpret data from large-scale and genome-wide expression studies. Oncomine (52) is a cancer microarray database that collects and curates 715 gene expression data-sets and 86,733 samples and associated clinical data from most major types of data. Oncomine allows a user to contact a gene-centric search to retrieve the differential expression analyses of a gene of interest across all available datasets; in a study-centric search, the genes that are differentially expressed in the selected study are provided. To facilitate data mining, the current version of Oncomine enables multi-gene search, gene ontology-based filtering and integration of Oncomine concepts. Integrated Tumor Transcriptome Array and Clinical data Analysis (ITTACA) (53) is a central repository of transcriptome microarray and associated clinical data from breast carcinoma, bladder carcinoma and uveal melanoma. A web interface offers different options for class comparison analyses, such as the comparison of profiles of expression distribution and patient survival analyses. The user is able to analyze the differential expression of one or more gene between two groups of samples with different phenotypes, and, conversely, the genes differentially expressed between two groups of samples (Fig. 3).

The Cancer Gene Expression Database (CGED) (54) contains cancer gene expression profiles and related clinical information. The expression data are obtained by adaptor-tagged competitive PCR from breast, colorectal, esophageal, gastric, hepatocellular, lung and thyroid cancers and glioma. The database can be queried either using gene identifiers or by functional categories. Mosaic plots are used for the visualization of gene expression data and comparison of the expression patterns of various genes. CancerMA (55) is an integrated bioinformatic pipeline used for the automated identification of novel candidate cancer biomarkers by analyzing the expression profiles of a user-defined gene list across public cancer microarray [GEO, ArrayExpress (56)] experimentally verified datasets. A total of 80 microarray datasets covering 13 types of cancer are available.

Differentially expressed proteins (DEPs) that contribute to the onset and progression of cancer have been identified. The first database of DEPs in human cancers, dbDEPC (57,58), currently contains 4,029 DEPs, curated from 331 mass spectrometry experiments across 20 types of human cancer. This resource enables the users to investigate whether a protein of interest has altered in particular cancers and, to create an association network of query proteins. Moreover, dbDEPC shows a heatmap that represents the expression profiles of a certain protein across various cancer types. An example of how to query dbDEPC is shown in Fig. 4.

The genes encoding protein kinases, enzymes that phosphorylate proteins, are among the most commonly mutated genes in human cancers. The MoKCa database (59) provides a collection of the mutations present in protein kinases involved in cancer, along with structural and functional annotation and, wherever possible, prediction of the impact of these mutations in the structure and function of kinases. The user can select from a pull-down list the gene that codes for a protein kinase: information is available for the types of mutations (e.g., missense, silent) found in tumor cell lines and, the mutated amino acid residues are mapped onto the tertiary structures of the affected protein kinase domains.

Cancers are thought to be initiated and maintained by a subpopulation of stem or stem-like cells with tumorigenic potential (60). SCDE (61,62) is an integrated repository of curated tissue and cancer stem cell data from blood, brain and intestine. The datasets are homogenized with regard to structure, formatting and annotation and stored in the Investigation/Study/Assay-Tab (ISA-Tab) format. SCDE is linked to the Galaxy framework which provides a series of analytical tools to compare those data to genes, molecular signatures and pathways. The CellLineNavigator (63) database contains gene expression profiles (generated uniformly) of >300 human cancer lines. On the basis of their phenotypic attributes, these cell lines were further classified into 28 tissues of origin and 57 different disease states. The database is also linked to advanced tools of bioinformatics analyses. The database can be searched for i) differentially expressed genes; ii) pathological or physiological states; and iii) gene names or functional characteristics, such as Gene Ontologies (GOs) (64) and KEGG pathway maps. A combination of all query options is also possible.

Chromosomal aberrations, such as translocations and their corresponding gene fusions and duplications or deletions and generated gene gains or losses are known to play an important role in the onset of tumorigenesis (65). The Mitelman Database of Chromosome Aberrations and Gene Fusions in Cancer (65) is a repository of chromosomal rearrangements such as translocations and their resulting gene fusions that are associated with tumor characteristics. The data contained in the database have been manually extracted from literature. The current version of the database includes a total of 64,679 cases and 2,094 gene fusions. The Mitelman Database can be queried by several different search options. In particular, the user can search for individual patient cases or associations of cases, such as concurrent chromosomal aberrations and clinical associations of cytogenetic abnormalities, or references themselves.

Tumor-associated antigens (TAAs) have been applied extensively in the clinical diagnosis and treatment of human cancers. The publicly available Human Potential Tumor Associated Antigen (HPtaa) (66) database contains potential TTAs identified by in silico computing. HPtaa incorporates publicly available microarray expression data, GEO’s SAGE data and Unigene (67) expression data, as well as other relevant knowledge bases such as CGAP. Currently, a total of 3,518 potential targets are included in the database. A web query interface enables users to search for potential TAAs overexpressed in several cancer types with particular gene features including chromosome (X and Y or euchromosome), coding capacity (protein-coding genes or else) and subcellular location (membrane or secretory proteins). The human immune response (humoral and cellular) to an increasing number of TAAs has also been well documented. The Academy of Cancer Immunology supported by the Ludwig Institute (both from New York, NY, USA) for Cancer Research have established the Cancer Immunome Database (CID) (68) which provides information on all the gene products against which an immunome response has been reported in cancer patients. The user can access information regarding the genes that encode the cancer antigens. Given that a gene can yield multiple antigenic epitopes, the frequency with which the antigenic epitopes are recognized by sera or cells from healthy and diseased individuals is reported. Wherever appropriate, access is provided to experimental evidence (serological results, microscope images and cytotoxic assays) or patients’ information (the type of cancer from which they are suffering, the disease stage, and time the samples were obtained). CTdatabase (69) is a curated repository of annotated and computationally predicted cancer-testis (CT) antigens. The CT antigens are broadly classified according to their expression pattern in human healthy tissue. CTdatabase also provides information on genes, the verified splice variants, genomic locations, gene duplications and bibliographical references.

Knowledge bases dedicated to cancer translational research and identification of drugs and compounds that inhibit cancer-related target genes are also available. CanSAR (70) is a public resource that supports cancer translational research and finding of drug through the integration of biological, chemical, pharmacological and disease data, structural biology and cellular networks. The user, through a single portal, is able to access information regarding genes, protein families, cell lines and compounds, as well as approved drugs and clinical candidates associates with cancer. CancerResource (71) is a comprehensive knowledge base that integrates cancer-relevant relationships of compounds/drugs and targets deduced from the text mining of >19 million PubMed abstracts and external resources such as Therapeutic Target Database (TTD) (72), Comparative Toxicogenomics Database (CTD) (73), Pharmacogenomics Knowledge Base (PharmGKB) (74) and DrugBank (75). CancerResource can be queried by Cancer (the user can view the genes expressed in specific cancer tissues, and also browse cancer-related KEGG pathways), Drug (the user can search by compound or drug and obtain information about the cancer relevance of the query drug/compound and its interactions with targets) and Target (drugs interacting with targets). The Anticancer Agent Mechanism Database (76,77) contains a list of 122 compounds with anticancer activity classified by their mechanism of action into alkylating agents, topoisomerase I/II inhibitors, RNA/DNA antimetabolites, and antimitotic agents. This set is generated by neural networks able to predict the mechanism of action of a drug based on its pattern of activity against a diverse panel of human tumor cell lines in the NCI drug screening program.

A major obstacle in cancer therapies is the development of drug resistance based on mutations in drug targets. Therefore, it is important to identify mutations in drug targets responsible for drug resistance. CancerDR (78) provides information of 148 anticancer drugs and their pharmacological profiling across ~1,000 cancer cell lines. Pharmacological profiling information of these anticancer drugs was collected from the Cancer Cell Line Encyclopedia (CCLE) (79) and COSMIC databases. CancerDR provides information about each drug target (cancer genes) that corresponds to these anti-cancer drugs, such as gene sequences in respective cancer cell lines, mutations, function and structure. This database allows users to search for drug targets, drugs, cell lines and structure. Clustering of cell lines on the basis of their drug sensitivity towards a drug target allows users to identify groups of cell lines, which are resistance to a particular anticancer drug, as well as multipotent drugs effective against a wide range of cancer cell lines. The clustering of sequences of a drug target is important to identify mutants/variants against the corresponding drug target.

IntOGen (80) is an integrative resource of high-throughput data associated with genomic, transcriptional, mutational alterations and modules (e.g., GO terms, KEGG pathways) involved in carcinogenesis. IntOGen collects data from various resources such as COSMIC, GEO, ArrayExpress, Progenetix (81), TCGA and CGP. Tumor samples in IntOGen are annotated with terms from the International Classification of Diseases for Oncology (ICD-O) (82) where the tumors are classified based on their topography (location in the human body) and histology (morphology). IntOGen can be queried by Genes, Projects, Cancer sites and Pathways. The Biomart portal (83) enables more complicated queries and the bulk download of all analysis results. The interface has a number of filters and attributes. IntOGen Biomart can be queried based on i) IntOGen Experiments, where the user can query gene or module (e.g., GO terms, KEGG pathways) information; ii) IntOGen Combinations, where the user is allowed to query a combination of experiments annotated with the same ICD-O term; and iii) IntOGenOncomodules which enables the user to search for combinations and experiments (Fig. 5). NCG 4.0 (84,85) is the current version of the Network of Cancer Genes, a repository of systems-level properties of cancer genes and oncomiRs (cancer-related microRNAs). It compiles information on 2,000 cancer genes that have been reported in literature to be mutated in 23 different types of cancer collected from 3,460 whole-exome and -genome screenings of cancer samples. NCG 4.0 reports information on the duplicability, functional annotation, evolutionary origin and interactions with other human proteins and microRNAs.