Inside the Cancer Gene Census: How COSMIC identifies cancer drivers
The Cancer Gene Census (CGC) is COSMIC’s authoritative list of genes with a causal role in cancer, categorised into two evidence-based tiers. Meticulously curated through rigorous scientific review, the CGC gives researchers, clinicians, and biotech teams a reliable way to identify cancer driver genes, prioritise targets, and interpret genomic data with confidence.
But how does a gene qualify for the CGC?
In this post, we’ll take you inside the expert-led curation process, from initial identification to eligibility checks and in-depth evidence reviews. You’ll see how COSMIC ensures every CGC gene is backed by robust, peer-reviewed science. This makes the CGC a dependable, time-saving foundation for cancer research and precision medicine.
Candidate Gene ID and ranking
The CGC review process begins by systematically identifying potential candidate genes from a wide range of sources. Our curators start by collating evidence from COSMIC’s own variant curation, large-scale cancer gene mutation analyses, and associations between genes and key cancer-relevant biological processes.
This integration of data helps us identify genes that may have a meaningful role in cancer development, on the basis that they are both frequently mutated in at least one cancer type, and their activity affects one, or more, of the Hallmarks of Cancer. These candidate genes are then assessed in greater detail, focusing on whether their somatic mutation profiles within COSMIC are consistent with that of a role as an oncogene, tumour suppressor gene.
This rigorous first step ensures that only the most promising candidates move forward, whilst grounding the entire process in robust, high-quality data.
Cancer Gene Census eligibility review
Following initial candidate ranking, shortlisted genes are subject to a targeted review to determine whether the gene is affected by recurrent somatic mutations or rare, pathogenic germline variants associated with cancer.
This involves a comprehensive examination of published studies reporting variant types and frequencies, and their disease contexts. If at least two independent studies provide compelling evidence, the gene advances to a deeper literature review focusing on functional impact, specifically, whether the wild-type gene-encoded protein has been experimentally shown to affect one or more hallmarks of cancer.
Regardless of whether a gene was initially delineated as a Tier 1 or Tier 2 candidate, progression to the final stage of a CGC eligibility review requires evidence from at least two peer-reviewed independent studies demonstrating functional relevance. This ensures only genes with robust empirical evidence are considered for inclusion.
Complete evidence review
Here, a gene’s candidacy is evaluated in its entirety. At this point, the central question is asked: Does this gene have a clearly definable role in cancer development and/or progression?
COSMIC’s team of post-doctoral scientists begins by collectively assessing the full body of evidence curated in the course of the review. Each gene is also evaluated for the conceptual feasibility, given a gene’s function, of its involvement in cancer development and/ or progression. Curators also consider the strength and quality of supporting studies to mitigate potential bias or overinterpretation.
Only after detailed discussions, and consensus among curators has taken place, is a gene formally accepted into the Cancer Gene Census and assigned a final Tier 1 or 2 classification, reflecting the strength and clarity of the supporting evidence.
Exemplary Annotated Data
Every gene in the Cancer Gene Census has undergone a meticulous, multi-stage review process led by expert curators and grounded in high-quality scientific evidence. For users, this means you can rely on the CGC to quickly identify well-validated cancer driver genes, reduce time spent evaluating conflicting studies, and confidently support research, clinical interpretation, and data analysis with an authoritative, up-to-date resource.
