A new approach for detecting low-level mutations in next-generation sequence data

We propose a new method that incorporates population re-sequencing data, distribution of reads, and strand bias in detecting low-level mutations. The method can accurately identify low-level mutations down to a level of 2.3%, with an average coverage of 500x, and with a false discovery rate of less than 1%. In addition, we also discuss other problems in detecting low-level mutations, including chimeric reads and sample cross-contamination, and provide possible solutions to them.