Search for a command to run...
The detection and quantification of genetic heterogeneity in populations of cells is fundamentally important to diverse fields, ranging from microbial evolution to human cancer genetics. However, despite the cost and throughput advances associated with massively parallel sequencing, it remains challenging to reliably detect mutations that are present at a low relative abundance in a given DNA sample. Here we describe smMIP, an assay that combines single molecule tagging with multiplex targeted capture to enable practical and highly sensitive detection of low-frequency or subclonal variation. To demonstrate the potential of the method, we simultaneously resequenced 33 clinically informative cancer genes in eight cell line and 45 clinical cancer samples. Single molecule tagging facilitated extremely accurate consensus calling, with an estimated per-base error rate of 8.4 × 10(-6) in cell lines and 2.6 × 10(-5) in clinical specimens. False-positive mutations in the single molecule consensus base-calls exhibited patterns predominantly consistent with DNA damage, including 8-oxo-guanine and spontaneous deamination of cytosine. Based on mixing experiments with cell line samples, sensitivity for mutations above 1% frequency was 83% with no false positives. At clinically informative sites, we identified seven low-frequency point mutations (0.2%-4.7%), including BRAF p.V600E (melanoma, 0.2% alternate allele frequency), KRAS p.G12V (lung, 0.6%), JAK2 p.V617F (melanoma, colon, two lung, 0.3%-1.4%), and NRAS p.Q61R (colon, 4.7%). We anticipate that smMIP will be broadly adoptable as a practical and effective method for accurately detecting low-frequency mutations in both research and clinical settings.