Accurate detection of somatic mutations with low variant allele fraction (VAF, <1%) has proven difficult. The lack of benchmarking datasets with such low-VAF mutations makes it more difficult to evaluate the applicability of conventional methods. Here, we have generated two different benchmarking datasets with multiple sequencing platforms, which contains somatic mutations with varying levels of VAF. We have also generated two different levels of replication data (sequencing- and library-level replicates) to test the effect of replication on detecting low-VAF variants. Followings are the detailed description of generated datasets.
1) Serially diluted gDNA mixture data
Genomic DNA from two independent blood samples was mixed to mimic somatic mutations at four different VAFs: 0.1%)> More...
Accurate detection of somatic mutations with low variant allele fraction (VAF, <1%) has proven difficult. The lack of benchmarking datasets with such low-VAF mutations makes it more difficult to evaluate the applicability of conventional methods. Here, we have generated two different benchmarking datasets with multiple sequencing platforms, which contains somatic mutations with varying levels of VAF. We have also generated two different levels of replication data (sequencing- and library-level replicates) to test the effect of replication on detecting low-VAF variants. Followings are the detailed description of generated datasets.
1) Serially diluted gDNA mixture data
Genomic DNA from two independent blood samples was mixed to mimic somatic mutations at four different VAFs: 0.5%, 1%, 5%, and 10%. Mixtures are then sequenced with three distinct platforms: hybridization-capture-based Illumina sequencing (ILH) and amplicon-based Illumina and Ion-Torrent sequencing (ILA and ITA, respectively). Sequencing libraries from each level were sequenced twice to generate data of sequencing replicates (designated as X12). For low-level mixtures with the VAF of 0.5% and 1%, two library-level replications were additionally prepared and sequenced (X21 and X31) to compare the effect of sequencing and library-level replication.
2) Commercial reference standard data
Tru-Q7 reference standard (1.3% Tier, HD734, Horizon Dx, Cambridge, UK) was sequenced with two different platforms: hybridization-capture-based Illumina sequencing (ILH) and amplicon-based Ion-Torrent sequencing (ITA). Widely-used cancer panels were used (SureSelect custom panel and Ion AmpliSeq cancer hotspot panel v2) to generate sequence data. Triplicate of library-level replication were made and sequenced with matched control (wild-type reference, HD752). Each library-level replicate data is designated as rep1, rep2, and, rep3, respectively. Less...