Detecting low-allele-fraction mosaic mutations: A multi-institution benchmarking with cell-line mixtures.
Abstract
Mosaic mutations in normal tissues occur at low variant allele fractions (VAFs), complicating detection. To benchmark variant-calling strategies, the SMaHT Network created a cell-line mixture (1:49) and produced ultra-deep short- and long-read whole-genome sequencing data across five centers (170×-470× per center). We assembled a reference set of mosaic SNVs and indels, cross-validated with long-read data to expose short-read limitations. We also partitioned the genome by mappability to examine the impact of genomic context, added a negative reference set, and accounted for culture-derived mutations. When seven institutions applied twelve pipelines, callsets were largely discordant across tools and replicates, partly reflecting stochastic sampling of low-VAF mutations in biological replicates. For SNVs with VAFs above 2%, sensitivity and precision approached 80% at ≥300×, with little gain from additional sequencing. This work establishes a framework and best practices for reliably detecting low-VAF mutations in non-cancer tissues and provides a valuable resource for the community.