Detecting rare AID-induced mutations in B-lineage oncogenes from High-Throughput Sequencing data using DeMinEr

Motivation: In B-lineage cells, the cytidine deaminase AID not only generates somatic mutations to variable regions of immunoglobulin genes but also inflicts, at a lower frequency, mutations to several non-Ig genes named AID off-targets that include proto-oncogenes. High-throughput sequencing (HTS) should be in principle the method of choice to detect and document these rare nucleotide substitutions. So far, HTS-based methods are impaired by a global sequencing error rate that usually covers the real mutation rate of AID off-target genes in activated B cells.Results: We demonstrate the validity of a per-base background subtraction method called DeMinEr (Detection of Minor variants by Error-correction) which uses deep-sequencing data from mutated and non- mutated samples to correct the substitution frequency at each nucleotide position along the sequenced region. Our method identifies somatic mutations at a frequency down to 0.2‰ at any nucleotide position within two off-target genes, Cd83 and Bcl6. Quantitative and qualitative validations demonstrate the accuracy and the sensitivity of DeMinEr when compared to standard methods. The high resolution and robustness of DeMinEr enable us to document fine effects such as age-dependent accumulation of mutations in these oncogenes.