04 · 21 proprietary layers
Layer 3 of 21 · Six shared core layers
SigPrint Mutational signatures and aetiology.
Why it's needed
What mutational patterns may be present, and what could they indicate in an appropriate evidence context?
What it draws on
1 of 8 evidence domains.
What it produces
Relevant signature observations and evidence-linked interpretation, where available.
Why it's essential
Layer 3 of 21.
1 · What it does
What SigPrint does
Processes that damage DNA, such as tobacco smoke, ultraviolet light, defective DNA repair or some earlier treatments, tend to leave characteristic patterns of mutations. SigPrint is intended to describe which of these patterns may be present in the tumour's mutations.
2 · Scientific question
The question it addresses
What mutational patterns may be present, and what could they indicate in an appropriate evidence context?
Intended function
Explain signature analysis and interpretation in context.
3 · Data inputs and required context
What it draws on
- 01 Molecular & genomic
Somatic mutation calls, ideally from a broad assay such as whole-exome sequencing.
4 · Scientific basis
The science behind it
Scientific basis
Signature analysis typically sorts single-base substitutions by their surrounding sequence context and fits them against reference signature catalogues. Reliable fitting needs enough mutations, which generally favours exome-scale data over small panels and tumours with a higher mutation count.
5 · Output and evidence sources
What it produces
Main output
Relevant signature observations and evidence-linked interpretation, where available.
Output type
- Computational inference. Estimated by analysis of available data rather than measured directly.
- Evidence-linked interpretation. Connects a finding to published or curated evidence, which has a source and a date.
6 · Why it\u2019s essential
How it powers the Twin
Provides supporting context rather than a standalone result. For example, a pattern associated with a DNA-repair defect may be read alongside HRD or MSI results, both of which are integrated analytical features of Twin.
Report section
7 · Why it\u2019s needed
What it changes — for doctors, patients and researchers
Clinicians
Supporting biological context to read alongside directly measured biomarkers.
Doctor & Clinical TeamPatients & caregivers
May help explain what contributed to the tumour's mutations. It is an association, not a verdict on cause.
Patient & Caregiver8 · Dependencies and limitations
Limitations
- Avoid causal certainty: signatures are associated with processes, not proof of a specific exposure in one person.
- Low mutation counts, common with targeted panels or low-TMB tumours, make signature estimates unreliable.
- Signature method, assay limitations and confidence should be reported only once established.
9 · Review and uncertainty
How the output is handled
Output status
An AI-assisted analytical output presented as structured findings for clinician review. It becomes part of a report only after the treating oncologist has reviewed and signed it.
How uncertainty is communicated
Signature contributions are reported with the number of mutations they rest on, and flagged as insufficient when the data cannot support an estimate.
10 · Claim boundary
What this layer does not establish
It does not prove what caused an individual's cancer.
The treating oncologist owns every decision. Every Twin report is reviewed and signed by the treating oncologist, then taken to KPCIRC's Molecular Tumour Board.
KPCIRC sign-off process