- Cancer is a genetic disease driven by the accumulation of genomic and epigenomic alterations that affect cellular growth, differentiation, survival, and DNA repair. These alterations may be inherited through the germline or acquired during a person’s lifetime in individual cells. Acquired alterations are known as somatic mutations and can contribute to tumour development, progression, metastasis, and resistance to treatment. Cancer mutation databases are essential bioinformatics resources for collecting, organizing, analysing, and interpreting these alterations across different cancer types. They enable researchers to identify recurrent mutations, distinguish potential driver events from passenger mutations, investigate cancer-associated genes, and explore molecular mechanisms underlying tumour development.
- Somatic mutations include single-nucleotide substitutions, small insertions and deletions, copy-number alterations, gene fusions, chromosomal rearrangements, and other structural variants. Cancer genomes may also contain changes in DNA methylation, gene expression, chromatin regulation, and other molecular features. The effects of these alterations depend on their genomic locations, affected genes, functional consequences, tumour type, and cellular context. Some mutations activate oncogenes, while others inactivate tumour-suppressor genes or disrupt DNA repair pathways. A single tumour can contain many mutations, but only a subset may contribute directly to cancer development or progression.
- The Catalogue Of Somatic Mutations In Cancer (COSMIC) is one of the major resources for studying somatic mutations in human cancer. Established in 2004 by the Wellcome Sanger Institute, COSMIC was developed to collect and standardize information about mutations reported in cancer research. It integrates information from scientific publications, large-scale cancer genome projects, targeted sequencing studies, and other datasets. Expert scientific curators review and organize published evidence, while computational pipelines support the processing and annotation of large genomic datasets. COSMIC has developed into a comprehensive cancer genomics knowledgebase containing information about small variants, structural alterations, gene fusions, copy-number changes, and other molecular features.
- COSMIC organizes cancer mutation information using several connected data components, including genes, variants, samples, cancer types, tissue origins, publications, and supporting evidence. A sample represents a biological specimen or cancer-derived material, while a mutation record describes an alteration detected or reported in that sample. Cancer type and tissue information help researchers compare mutation patterns across tumour groups. Publication references provide links to the scientific evidence from which curated information was obtained. Standardized gene identifiers, genomic coordinates, variant descriptions, and cancer classifications make it possible to integrate COSMIC data with other genomic and clinical resources.
- The database includes information about several major classes of cancer-associated genomic alterations. Single-nucleotide variants involve changes to individual DNA bases and may produce missense, nonsense, synonymous, or non-coding effects. Insertions and deletions can alter protein sequences, introduce premature stop codons, or disrupt reading frames. Copy-number alterations involve the gain or loss of genomic regions and can lead to oncogene amplification or tumour-suppressor gene deletion. Structural rearrangements may produce gene fusions or alter regulatory relationships between genomic regions. Gene expression and DNA methylation information can provide additional context for understanding molecular changes in cancer.
- COSMIC includes several specialized resources and modules that extend its usefulness beyond a catalogue of individual mutations. The Cancer Gene Census (CGC) identifies genes for which substantial evidence supports a role in cancer. It classifies and describes cancer-associated genes according to their established biological roles and mechanisms of dysfunction. Cancer genes commonly include oncogenes, tumour-suppressor genes, and genes involved in DNA repair, cell-cycle regulation, chromatin organization, and other processes. The Cancer Gene Census is useful for identifying genes with established cancer relevance, although inclusion does not mean that every variant in a listed gene is oncogenic or clinically actionable.
- The Cancer Mutation Census (CMC) focuses on the interpretation of mutations in cancer-associated genes by collecting evidence about the functional and biological significance of specific alterations. It helps connect individual mutations with their effects on proteins and their potential roles in cancer. This is important because different mutations within the same gene can have different consequences. For example, one alteration may activate an oncogene, another may have no measurable functional effect, and a third may disrupt an important protein domain. Mutation-level interpretation therefore requires more detailed evidence than simply identifying a gene as cancer-associated.
- The COSMIC Mutational Signatures resource describes characteristic patterns of mutations associated with different biological processes and mutational mechanisms. A mutational signature is a pattern of mutation types and sequence contexts observed across a cancer genome. Such patterns may reflect exposure to environmental mutagens, defects in DNA repair, activity of specific enzymes, or other processes that generate DNA damage. Single-base substitution signatures, doublet-base substitution signatures, insertion and deletion signatures, copy-number signatures, and structural-variation signatures can provide information about the processes that have shaped a tumour genome. Signature analysis can help investigate cancer aetiology, identify DNA repair deficiencies, and explore possible therapeutic vulnerabilities. However, signature assignment depends on data quality, the number of mutations available, and the statistical method used.
- The COSMIC Cell Lines Project provides molecular information about cancer cell lines commonly used in laboratory research. Cell lines are important experimental models for studying cancer biology, testing drugs, investigating molecular pathways, and validating computational predictions. The project includes genomic and molecular profiles such as sequence variants, copy-number alterations, gene expression, RNA sequencing, and DNA methylation information for selected cell lines. Researchers can use these data to identify cell lines carrying particular mutations or molecular characteristics and select appropriate experimental models. Cell lines do not fully reproduce the complexity of human tumours, however, and their molecular properties may change during long-term culture.
- Other COSMIC modules provide information about structural and functional consequences of cancer mutations. COSMIC-3D connects mutations with protein structures, allowing researchers to investigate their positions in three-dimensional molecular models and explore possible effects on protein function or interactions. Resistance-related information records mutations associated with reduced sensitivity or resistance to particular cancer therapies. The Actionability resource connects genomic alterations with therapeutic information, clinical trials, and treatment-related evidence. These resources help bridge cancer genomics and translational research, but their annotations should be interpreted in the context of the evidence, tumour type, treatment setting, and current clinical guidance.
- Cancer mutation databases are not limited to COSMIC. Several complementary resources provide different types of genomic, clinical, functional, and population-level information. The Cancer Genome Atlas (TCGA) is a major cancer genomics initiative that generated multi-omics datasets across numerous tumour types. It includes genomic mutations, copy-number alterations, gene expression, DNA methylation, protein measurements, and clinical information. TCGA has supported large-scale studies of cancer molecular subtypes, tumour classification, driver alterations, and relationships between genomic changes and clinical outcomes. Its data are widely reused in cancer bioinformatics and are accessible through portals such as the Genomic Data Commons.
- The International Cancer Genome Consortium (ICGC) coordinated large-scale studies of genomic alterations across many cancer types and populations. Its projects generated whole-genome, whole-exome, transcriptomic, and other molecular datasets, contributing to the understanding of cancer diversity and molecular mechanisms. The ICGC and its associated resources provide valuable information for comparing tumour genomes across studies and populations. The ICGC ARGO initiative extends this international effort by emphasizing clinical and genomic data integration to support precision oncology. Researchers should check the relevant project and data-access policies because availability and access conditions differ among datasets.
- cBioPortal for Cancer Genomics is an important platform for exploring and visualizing large cancer genomics datasets. It integrates molecular profiles and clinical information from studies such as TCGA and other cancer research projects. Users can investigate mutations, copy-number changes, gene expression, survival information, and relationships between molecular alterations and clinical variables. cBioPortal supports queries involving individual genes, gene sets, and patient cohorts, making it useful for exploratory analysis and hypothesis generation. Its visualizations can help identify recurrent alterations, co-occurring mutations, mutually exclusive events, and molecular patterns across tumour groups.
- The AACR Project GENIE (Genomics Evidence Neoplasia Information Exchange) is a large-scale international cancer genomics data-sharing initiative that combines genomic profiling information from patients treated at participating cancer centres. It provides clinically generated sequencing data linked to selected clinical and demographic information, subject to data-use and privacy controls. GENIE is particularly valuable for studying genomic alterations in real-world cancer populations and examining molecular patterns across different institutions and tumour types. Its data may differ from research cohorts because the patients and samples were selected within clinical testing environments rather than through uniform population-based sampling.
- CIViC (Clinical Interpretation of Variants in Cancer) is an open knowledgebase focused on the clinical interpretation of cancer variants. It organizes evidence linking genomic alterations with clinical significance, including diagnostic, prognostic, and predictive implications. Evidence items are associated with specific molecular profiles, diseases, therapies, and supporting publications. CIViC is useful for understanding how a cancer-associated variant may influence clinical interpretation, but its evidence must be assessed for strength, relevance, and currency before use in clinical contexts.
- OncoKB is a precision oncology knowledgebase that connects cancer-associated genetic alterations with biological and clinical evidence, including therapeutic implications. It provides information about alterations that may be relevant to treatment selection and resistance, with evidence levels reflecting the strength and clinical context of the available data. Cancer Genome Interpreter (CGI) is another resource designed to help interpret genomic alterations by connecting them with cancer driver evidence and potential therapeutic associations. These resources complement mutation archives by focusing more directly on functional and clinical interpretation rather than simply recording the presence of a mutation.
- Other specialized cancer mutation resources include IntOGen, which identifies and prioritizes cancer driver genes and mutations through computational analysis of large cancer genomics datasets; cancerhotspots.org, which helps identify recurrent mutation hotspots in cancer-associated proteins; and DepMap, which connects molecular profiles of cancer cell lines with genetic dependency and drug-response information. The Cancer Dependency Map supports investigations into genes and pathways required for cancer cell survival, while pharmacogenomic resources can help relate molecular alterations to drug sensitivity. These datasets are particularly useful for identifying candidate therapeutic targets and designing functional experiments.
- The analysis of cancer mutations also relies on general genomic variation and clinical annotation databases. ClinVar contains submitted interpretations of the clinical significance of genetic variants, including some variants relevant to hereditary cancer predisposition. ClinGen evaluates gene–disease relationships and develops expert guidance for clinical genomic interpretation. gnomAD provides population allele-frequency information that can help determine whether a variant is common in the general population or unusually rare. dbSNP and the European Variation Archive (EVA) provide archival information about genetic variants, while dbVar stores structural-variation observations. These resources have different scopes and should not be treated as interchangeable. In particular, a population-frequency database does not establish whether a somatic mutation drives cancer, and a clinical germline classification does not automatically apply to the same variant detected in a tumour.
- Cancer mutation databases support several important bioinformatics applications. One major application is the identification of potential driver mutations. Driver mutations provide a selective advantage to cancer cells by promoting proliferation, survival, invasion, or other tumour-associated characteristics. Passenger mutations accumulate during tumour development but do not necessarily contribute directly to cancer progression. Computational methods compare mutation frequencies, recurrence patterns, functional consequences, evolutionary selection, and other evidence to identify candidate drivers. Recurrent mutations in genes such as TP53, KRAS, EGFR, BRAF, PIK3CA, and APC have been extensively studied, although their biological and clinical significance depends on the exact alteration and cancer context.
- Another important application is the identification of cancer-associated pathways. Mutations affecting different genes may disrupt the same biological pathway, such as the MAPK signalling pathway, PI3K–AKT pathway, Wnt signalling pathway, p53 pathway, cell-cycle regulation, or DNA damage response. Pathway-based analysis helps researchers move beyond individual genes and understand how groups of molecular alterations contribute to tumour development. Databases such as COSMIC, Reactome, KEGG, Gene Ontology, and pathway interaction resources can be combined to investigate these relationships. However, finding mutations in a pathway does not necessarily demonstrate that the pathway is activated or inhibited; functional evidence is required to establish the direction and biological consequences of the changes.
- Cancer mutation databases are also important in precision oncology. Tumours from different patients, even when they originate in the same organ, may contain distinct molecular alterations. Identifying these alterations can help classify tumours into molecular subgroups and identify potential treatment options. For example, certain EGFR mutations may be relevant to targeted therapy in specific lung cancers, while BRAF V600 alterations may be relevant in selected melanoma and other cancer settings. Other alterations may be associated with resistance to treatment or may indicate eligibility for clinical trials. Database annotations can support the investigation, but treatment decisions require validated diagnostic testing, current clinical guidelines, and assessment by qualified healthcare professionals.
- The study of mutational signatures is another major application. Cancer genomes contain accumulated mutations that reflect the biological processes operating during tumour development. Bioinformatics tools can identify these patterns by decomposing mutation profiles into known or newly inferred signatures. COSMIC signature definitions are widely used in this work, while tools such as SigProfiler provide methods for signature extraction, attribution, and analysis. Mutational signatures can reveal processes such as ultraviolet radiation exposure, tobacco-associated mutagenesis, defective DNA mismatch repair, homologous recombination deficiency, and activity of APOBEC enzymes. The presence of a signature may suggest a biological mechanism, but it should be interpreted with appropriate statistical support and independent biological evidence.
- Cancer mutation databases also support comparative cancer genomics. Researchers can compare mutation frequencies among tumour types, investigate differences between primary and metastatic cancers, and study the evolution of tumours over time. Multi-region sequencing and longitudinal sampling can reveal subclonal diversity and changes in mutation patterns during disease progression or treatment. The integration of genomic data with transcriptomic, epigenomic, proteomic, and clinical information provides a more complete picture of tumour biology. Such analyses are important for understanding tumour heterogeneity, metastasis, treatment response, and acquired drug resistance.
- Cancer mutation data are commonly represented using standardized formats and identifiers. The Variant Call Format (VCF) is widely used for genomic variant calls, while tab-separated files are common in database downloads. Variant descriptions may use genomic coordinates, reference and alternate alleles, transcript identifiers, protein consequences, and standardized nomenclature such as Human Genome Variation Society (HGVS) expressions. Genome assembly information is essential because coordinates differ between GRCh37 and GRCh38. COSMIC provides data mapped to supported genome assemblies and uses stable identifiers to improve the traceability and interoperability of somatic variant records. Researchers should retain the original database identifiers and release information when integrating data into analysis pipelines.
- Programmatic access is important for large-scale cancer mutation analysis. COSMIC provides web-based exploration and downloadable datasets, including options for browser-based, scripted, and filtered downloads. Access to data downloads is subject to registration and licensing conditions; non-commercial and commercial use may be governed by different terms. Other cancer resources provide web APIs, downloadable files, or analysis portals. Python and R can be used to retrieve, parse, filter, annotate, and visualize cancer mutation data. Tools such as Ensembl Variant Effect Predictor (VEP), ANNOVAR, and SnpEff can help annotate variant consequences, while Bioconductor packages and specialized cancer genomics tools support downstream analysis. Researchers should consult each resource’s current documentation and access conditions before building automated pipelines.
- Data integration requires careful attention to sample identifiers, tumour classifications, gene nomenclature, reference assemblies, and variant representation. The same mutation may be recorded under different identifiers or genomic coordinate systems in separate databases. A variant may also be reported at the DNA, transcript, or protein level, requiring normalization before records can be compared. Cancer type terminology may differ between publications and databases, and sample metadata may be incomplete. Reliable integration therefore requires consistent annotation, controlled vocabularies, standardized identifiers, and traceable links to the original evidence.
- Despite their importance, cancer mutation databases have limitations. The number of reported mutations is influenced by the availability of sequencing studies, publication practices, sample sizes, cancer prevalence, and research priorities. Some cancer types and patient populations are much better represented than others. Differences in sequencing platforms, variant-calling pipelines, tumour purity, sequencing depth, and quality-control procedures can affect which mutations are detected. A mutation absent from a database may simply have not been investigated or reported. Similarly, a mutation appearing frequently in a database may reflect the overrepresentation of particular tumour types or heavily studied genes rather than a universal cancer mechanism.
- A further challenge is distinguishing somatic mutations from germline variants. Tumour sequencing can detect both inherited and acquired alterations, and matched normal tissue may be needed to distinguish them reliably. Tumour-only sequencing can lead to uncertainty about variant origin, particularly for rare germline variants or variants affected by clonal haematopoiesis. Variant allele frequency alone is not sufficient to determine whether an alteration is somatic, because tumour purity, local copy number, subclonality, and normal-cell contamination influence the observed value. Correct interpretation requires consideration of sample design and supporting evidence.
- The distinction between a recurrent mutation, a driver mutation, and a clinically actionable mutation is also essential. A recurrent alteration may be biologically important, but recurrence alone does not prove that it drives tumour development. A driver mutation may contribute to cancer biology without having an established treatment implication. A clinically actionable alteration requires evidence that it can inform a diagnostic, prognostic, predictive, or therapeutic decision in a defined clinical context. Cancer mutation databases provide valuable evidence for these assessments, but they do not replace functional validation, clinical trials, expert review, or current clinical guidelines.
- Reproducible cancer genomics research requires documentation of database names, release versions, access dates, reference genome assemblies, filtering criteria, software versions, and analytical parameters. Researchers should preserve sample-level provenance where permitted, distinguish curated evidence from computational predictions, and record whether an annotation is based on a primary publication, a large-scale dataset, or an expert interpretation. Licensing and data-sharing restrictions must also be respected, particularly when working with patient-level genomic and clinical information. Genomic data can contain sensitive information, and access, storage, and reuse should follow applicable ethical, legal, and institutional requirements.
- Overall, COSMIC and other cancer mutation databases provide essential foundations for modern cancer bioinformatics. COSMIC offers expert-curated information about somatic mutations and their cancer context, while resources such as TCGA, ICGC, cBioPortal, AACR Project GENIE, CIViC, OncoKB, IntOGen, and DepMap contribute complementary genomic, clinical, functional, and therapeutic evidence. Their integration supports cancer gene discovery, driver mutation analysis, molecular classification, mutational signature research, drug-response investigation, and precision oncology. Effective use of these resources depends on understanding their scope, evidence standards, data representations, licensing conditions, and limitations. Together, they help transform large-scale cancer sequencing data into biological knowledge and clinically relevant insights.