Glossary

Mutational Signature

A characteristic pattern of nucleotide substitution contexts imprinted on a tumor genome by specific mutagenic processes, computationally deconvolved to identify etiological drivers.

Get in touch Learn more

Engineer optimizing context window usage on laptop, token usage charts visible, technical work session.

GENOMIC FINGERPRINT

What is a Mutational Signature?

A mutational signature is a characteristic pattern of nucleotide substitution contexts imprinted on a tumor genome by specific mutagenic processes, computationally deconvolved to identify etiological drivers.

A mutational signature is a distinctive combination of mutation types—defined by the base change and its immediate flanking bases—that arises from a specific DNA damage and repair process. Using non-negative matrix factorization, computational pipelines decompose a tumor's catalog of somatic mutations into constituent signatures, quantifying the contribution of each operative mutational process, such as ultraviolet light exposure or defective homologous recombination.

Each signature is mathematically represented as a probability vector across 96 trinucleotide substitution contexts, reflecting the mechanistic preferences of the underlying mutagen. Cataloged in the COSMIC Mutational Signatures database, these patterns serve as retrospective dosimeters, revealing past exposures and endogenous repair deficiencies. In clinical oncology, the presence of specific signatures, like the BRCA-associated Signature 3, directly informs therapeutic decisions regarding PARP inhibitor sensitivity.

DECONVOLVING THE ETIOLOGY OF CANCER

Core Characteristics of Mutational Signatures

Mutational signatures are the indelible fingerprints left on the genome by distinct DNA-damaging and repair processes. Computational deconvolution of these patterns reveals the chronological history of mutagenic exposures that drove a tumor's development.

The Trinucleotide Context

A mutational signature is defined not just by the base change (e.g., C>A) but by the immediate 5' and 3' sequence context. This creates 96 distinct mutation types (6 substitution types × 4 5' bases × 4 3' bases). The relative frequency of each type forms a characteristic profile. For example, UV radiation predominantly causes C>T transitions at dipyrimidine sites (TpC or CpC contexts), while tobacco carcinogens preferentially target G bases flanked by pyrimidines.

Mutation Classes

Non-Negative Matrix Factorization (NMF)

The core mathematical framework for extracting signatures from a catalog of somatic mutations. NMF decomposes a matrix of mutation counts (samples × 96 contexts) into two lower-rank matrices: signatures (processes × 96 contexts) and exposures (samples × processes). The algorithm iteratively optimizes these matrices to minimize reconstruction error while enforcing non-negativity, ensuring biologically interpretable results. This blind source separation identifies the latent mutagenic processes active across a cohort.

EXPLORE

Single-Base Substitution (SBS) Signatures

The most well-characterized class, cataloged in the COSMIC database. Each SBS signature is linked to a specific etiology:

SBS1: A clock-like signature from spontaneous deamination of 5-methylcytosine, correlating with chronological age.
SBS4: Strongly associated with tobacco smoking, characterized by C>A transversions.
SBS7a-d: Ultraviolet light exposure, dominated by C>T transitions at dipyrimidines.
SBS10a/b: Defective polymerase epsilon (POLE) proofreading, producing an enormous number of mutations.

80+

SBS Signatures Cataloged

Doublet Base Substitutions (DBS)

Two consecutive nucleotide changes occurring simultaneously as a single mutational event. DBS signatures provide orthogonal evidence for specific mutagens. DBS2, characterized by CC>TT tandem mutations, is a highly specific hallmark of UV radiation damage. DBS4, featuring GC>TT substitutions, is linked to platinum-based chemotherapy agents like cisplatin, which form intrastrand crosslinks. Analyzing DBS patterns helps resolve ambiguous SBS assignments.

Insertion-Deletion (ID) Signatures

Small insertions and deletions (indels) are categorized by length, sequence composition, and the presence of flanking microhomology or tandem repeats. ID1 features long deletions with flanking microhomology, indicative of non-homologous end joining (NHEJ) repair deficiency. ID2 shows short deletions at repeat units, linked to DNA polymerase slippage. ID6, characterized by 1bp insertions at poly-T tracts, correlates with defective mismatch repair (dMMR) and microsatellite instability.

Signature Refitting & Attribution

Once reference signatures are defined, the exposure of a single tumor can be estimated by refitting—solving for the linear combination of known signatures that best reconstructs the observed mutation profile. Algorithms like SigProfilerAssignment or mmsig use constrained optimization to assign mutations to specific processes. This allows clinicians to quantify the contribution of a known carcinogen (e.g., UV light) or a defective repair pathway (e.g., BRCA1/2 deficiency) in an individual patient's tumor.

EXPLORE

MUTATIONAL PROCESS DECONVOLUTION

Frequently Asked Questions About Mutational Signatures

Explore the computational frameworks used to decipher the historical imprints of mutagenic processes on cancer genomes, from single-base substitution patterns to the mathematical algorithms that separate overlapping signals.

A mutational signature is a characteristic pattern of somatic mutations imprinted on a tumor genome by a specific mutagenic process, defined mathematically by the relative frequencies of mutation types within their trinucleotide sequence context. The standard catalog uses 96 mutational features, representing six classes of single-base substitutions (C>A, C>G, C>T, T>A, T>C, T>G) multiplied by the 16 possible combinations of the 5' and 3' flanking bases. Each signature is represented as a discrete probability vector over these 96 channels, normalized to sum to 1.0. This representation captures the mechanistic preferences of distinct mutagens—for example, ultraviolet light predominantly induces C>T transitions at dipyrimidine sites (TpCpN and CpCpN contexts), while aristolochic acid produces a distinctive T>A transversion signature enriched at CpTpG trinucleotides. The concept was formalized by the Catalogue of Somatic Mutations in Cancer (COSMIC) and the Wellcome Sanger Institute's computational framework, which established that the mutational landscape of most cancers can be decomposed into a linear combination of a finite set of underlying signatures, each reflecting a distinct DNA damage, repair deficiency, or enzymatic editing process.

COMPARATIVE ONCOGENOMICS

Mutational Signatures vs. Driver Mutations vs. Mutational Burden

Distinguishing three core concepts in cancer genomics that are often conflated: the etiological fingerprint of mutagenic processes, the functional consequence of specific alterations, and the quantitative load of genomic instability.

Feature	Mutational Signature	Driver Mutation	Mutational Burden
Primary Definition	A characteristic pattern of nucleotide substitution contexts imprinted by a specific mutagenic process	A somatic alteration that confers a selective growth advantage to the tumor cell	A quantitative measure of the total number of somatic coding mutations per megabase of tumor genome
Biological Role	Etiological: Reveals the historical mutagenic exposures and endogenous processes that shaped the genome	Functional: Directly activates oncogenes or inactivates tumor suppressors to drive clonal expansion	Quantitative: Measures the overall genomic instability and neoantigen potential of the tumor
Computational Detection	Non-negative matrix factorization (NMF) deconvolution of mutational catalogs into constituent signatures	Recurrence analysis, functional impact prediction (SIFT, PolyPhen), and clustering in mutational hotspots	Simple counting of non-synonymous coding variants per megabase from targeted or whole-exome sequencing
Input Data Requirement	Aggregated single-base substitution (SBS), doublet-base substitution (DBS), and indel mutational catalogs	Somatic variant callset with functional annotation against reference transcriptome	Somatic variant callset filtered for coding regions with sufficient coverage breadth
Clinical Utility	Retrospective: Identifies carcinogenic exposures (e.g., UV light, aristolochic acid) and DNA repair deficiencies (e.g., BRCAness)	Prognostic and Therapeutic: Directly targetable by small-molecule inhibitors (e.g., EGFR, BRAF) or indicates resistance	Predictive Biomarker: FDA-approved companion diagnostic for immune checkpoint inhibitor eligibility (e.g., pembrolizumab for TMB-H tumors)
Temporal Context	Historical: Represents the cumulative mutagenic history of the tumor cell lineage prior to sampling	Ongoing: The specific alteration is actively maintaining the malignant phenotype at time of sampling	Snapshot: A static count of mutations at the time of biopsy; does not distinguish historical from ongoing mutagenesis
Example	COSMIC Signature 3 (BRCA1/2 deficiency) or Signature 7 (UV radiation)	BRAF V600E in melanoma or EGFR L858R in lung adenocarcinoma	TMB-H defined as ≥10 mutations per megabase in solid tumors
Key Limitation	Cannot distinguish passenger mutations from drivers; requires large cohorts for robust deconvolution	Not all tumors harbor actionable drivers; intratumoral heterogeneity can confound single-biopsy analysis	Does not identify specific mutagenic processes; TMB alone is an imperfect predictor of immunotherapy response

Enabling Efficiency, Speed & Accuracy

Intelligent Analysis, Decision & Execution

We build AI systems for teams that need search across company data, workflow automation across tools, or AI features inside products and internal software.

Talk to Us

Search across company data

Give teams answers from docs, tickets, runbooks, and product data with sources and permissions.

Useful when people spend too long searching or get different answers from different systems.

Enterprise searchRAGPermissions

Automate internal workflows

Use AI to route work, draft outputs, trigger actions, and keep approvals and logs in place.

Useful when repetitive work moves across multiple tools and teams.

AI agentsWorkflow automationGovernance

Add AI to products and internal tools

Build assistants, guided actions, or decision support into the software your team or customers already use.

Useful when AI needs to be part of the product, not a separate tool.

AI integrationDecision supportModel routing

ETIOLOGICAL DECONVOLUTION