Cancer Cell Fraction
Cancer cell fraction (CCF) is the proportion of cancer cells in a sample that carry a specific somatic mutation. It is the fundamental metric for clone abundance in subclonal-reconstruction.
Calculation
CCF is derived from the observed variant-allele-fraction by correcting for sample purity (ρ) and local copy number:
Where:
| Symbol | Name | Meaning | How obtained |
|---|---|---|---|
| VAF | Variant allele fraction | Proportion of reads carrying the variant | Directly observed from sequencing |
| ρ | Tumor purity | Proportion of cancer cells in the sample | Estimated from sequencing (ABSOLUTE, ASCAT) or histology |
| N_T | Total copy number | Total DNA copies at the locus in tumor cells | CNA calling from logR + BAF segmentation |
| m | Multiplicity | How many of the N_T copies carry the SNV | Unknown in bulk sequencing; assumed = 1 |
| 2 × (1−ρ) | Normal contamination | Contribution of normal diploid cells to read depth | Derived from purity |
The critical distinction: N_T vs. m. N_T is a property of the genomic region — every SNV in a 1q-gain region shares N_T = 3. m is a property of the individual mutation — an SNV that predates the gain may have m = 2 or 3 (amplified along with the region); an SNV that arose after the gain has m = 1. Both SNVs share the same N_T = 3, but their true CCFs differ by a factor of m. The formula has one equation and two unknowns (CCF and m) whenever N_T ≠ 2 — this is the root cause of why SNVs in CNA regions are discarded from subclonal reconstruction (copy-number-alteration §5).
Clonal vs. Subclonal
A mutation with CCF = 1.0 (or close to it, after accounting for uncertainty) is clonal — present in 100% of cancer cells. It is likely a truncal mutation that occurred early in the tumor’s evolutionary history, before the most recent common ancestor of the sampled cells. A mutation with CCF < 1.0 is subclonal — present in only a fraction of cancer cells. It occurred later, after the divergence of the sampled population into multiple subclonal lineages.
CCF and Phylogenetics
The CCFs of different mutations constrain the possible phylogenetic-tree. A mutation present in all cells of clone A cannot have a higher CCF than a mutation present in all cells of clone A’s ancestor. In multi-sample data, the crossing-rule applies: if clone A has higher CCF than clone B in one sample but lower in another, they must be sibling clones, not ancestor and descendant.
Detection Limits
At standard sequencing depths (~100×), mutations with CCF < 0.05–0.10 are typically undetectable, meaning minor subclones below this threshold are invisible to bulk sequencing (Tarabichi et al., 2021).
Revision history
- 2026-07-12 — Expanded Calculation section with N_T vs. m distinction table (five-row variable glossary). N_T is a regional property (shared by all SNVs in the region); m is a mutation-specific property (varies by SNV depending on timing relative to CNA). This is the root cause of why SNVs in CNA regions are underdetermined. (copy-number-alteration)