LFMD: detecting low-frequency mutations in genome sequencing data without molecular tags

As next-generation sequencing (NGS) and liquid biopsy become more prevalent in clinic and research, for cancer diagnosis, molecular target identification, and disease monitoring, there is an increasing need for better methods to reduce cost and improve sensitivity and specificity of mutation detection. Since NGS has an error rate of around 1%, it is difficult to accurately and efficiently identify mutations with less than 1% frequency in a sample. Here we propose a likelihood-based approach, called Low-Frequency Mutation Detector (LFMD), which combines the advantages of duplex sequencing (DS) and bottleneck sequencing system (BotSeqS) to maximize the utilization of duplicate sequence reads. Compared with DS, the new method achieves higher sensitivity (improved by ~16%), higher specificity and lower cost (reduced by ~70%) without involving additional experimental steps, customized adapters or molecular tags. In addition, this method can also be used to improve sensitivity and specificity of other variant calling algorithms by making it unnecessary to remove polymerase chain reaction (PCR) duplicates.