Combination of RNA- and exome sequencing: Increasing specificity for identification of somatic point mutations and indels in acute leukaemia

Handling of large data sets from whole exome sequencing is a challenge, not the least because of the high risk of detecting false positive variants. Furthermore, there is still no consensus regarding what stage filtering of variants is performed in the bioinformatics pipeline to produce consistent result sets, the extent of filtering and how this is most optimal performed. We hypothesized that combination of coding transcriptome and exomes enables precise identification of both somatic single nucleotide and indel variants early in the bioinformatics process, superseding the need for extensive annotation and validation from external databases. Exome and RNA-sequencing were performed on diagnosis-remission paired bone marrow samples from 5 cytogenetically normal AML, i.e. sequencing of 20 samples in total. Intersection of rare exome and RNA variants, exclusively found in the diagnostic samples, yielded a median of 6 somatic point mutations and small indels, ranging from 3 to 21. Close correlation between routine diagnostic biomarkers was observed in 29/30 laboratory tests, with the exception of a large FLT3 internal tandem repeat, whereas WT1 with nonsense mutation lacked RNA transcripts. Additionally, the assay revealed mutations in DNMT3A, IDH2, TET2 and IL32 frame shift, not encompassed by routine panel. We thus describe a procedure to effectively reduce observations to a focused subset of high specificity. The approach is applicable to precise screening of both putative driver mutations and, often heterogeneous, discrete patient specific somatic combinations.