Accurate detection of sub-1% frequency somatic mutations by whole-genome sequencing.

Positive (iPSC, 0.1-25% VAF) and negative (fibroblast) control mutations; discovery with Illumina short reads (~1,360×), confirmation with combined PacBio and AVITI (~600×)

The authors assess high-coverage bulk whole-genome sequencing (WGS) for low-frequency somatic mutation detection, using primary human fibroblasts with naturally occurring mutations and expanded subclones, iPSC lines corresponding to the subclones and additional iPSC lines from other fibroblasts of the same donor. Mutations found in iPSC lines defined positive controls (0.1% to 25% variant allele fraction, VAF) and negative controls in fibroblasts; PacBio and AVITI were the most accurate platforms in easy (short-read accessible) and non-easy regions, respectively, with AVITI also most accurate for indels. In easy regions, using ~1,360× Illumina short-read data for discovery and ~600× combined PacBio and AVITI data for confirmation, point mutations were detected accurately and specifically below 1% VAF and non-homopolymer indels down to 1% VAF, whereas homopolymer indels were reliable only above 3% VAF. Detection was degraded in postmortem samples, for which additional filtering is recommended, and combining ~200× coverage across tissues from one donor proved feasible for detecting shared developmental mutations.

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

The thresholds are directly useful when searching for low-level mosaicism: below 1% VAF, only substitutions and non-homopolymer indels are reliable, while homopolymer indels remain out of reach below 3%. The study relies on fibroblasts and iPSCs from a single donor, with no patients, and on depths (~1,360×) far above those of a clinical genome. Since the benchmark covers only easy regions, these performances do not apply to regions inaccessible to short reads.

Analysis by Dr Thibaut Benquey