Initial PI3K assay application of this approach was performed on the somatic substitutions derived from the whole genomes of 21 breast cancer patients [33••]. In order to increase the resolution of the derived mutational signatures, substitutions

were examined using their immediate sequencing context. This included the base immediately 5′ before the somatic mutation and the base immediately 3′ after the somatic mutation; thus resulting in 96 mutation types — 16 different for each of the six types of somatic substitutions. For example, C > T mutations were extended to include C > T with (5′ adenine): ApCpA, ApCpC, ApCpG, ApCpT; (5′ cytosine): CpCpA, CpCpC, CpCpG, CpCpT; (5′ guanine): GpCpA, GpCpC, GpCpG, GpCpT; and (5′ thymine): TpCpA, TpCpC, TpCpG, TpCpT. Including the immediate sequence context allows better differentiation between different mutational processes; for example, distinguishing between C > T mutations due to the formation UV-light induced photodimers (i.e. C > T mutations at dipyrimidine sites such as TpCpC or CpCpC) from C > T mutations due to deamination of 5-methylcytosine (i.e. GDC-0980 C > T mutations at CpG sites). The mutational catalogues of the 21 breast cancer genomes were generated,

including each of the 96 mutation types, and applying the newly developed method to these catalogues revealed multiple distinct mutational signatures of substitutions. As expected, a mutational signature almost with features of C > T mutations at CpG sites was identified in most samples, thus reflecting the activity of normal endogenous cellular processes. Further, a mutational

signature with C > X mutations at TpC sites was identified and based on similarity between its mutational pattern and in vivo experimental data, it was proposed that this process is due to the activity of the APOBEC family of deaminases and more specifically APOBEC1, APOBEC3A, and/or APOBEC3B [ 84 and 85]. Additionally, a rather uniform mutational signature (no prominent features across trinucleotides) was also identified and, interestingly, the activity of this mutational signature in each of the 21 samples allowed separation (by unsupervised hierarchical clustering) of BRCA1 and BRCA2 wild-type breast tumours from BRCA1 and BRCA2 germline mutants. Another mutational signature with unknown aetiology and mutations predominately at C > G at TpC was also identified. In addition to these genome-wide signatures, a localized hypermutation (termed kataegis) was observed in some of the breast cancer samples. This localized hypermutation was predominantly constituted of C > T and C > G substitutions at TpC trinucleotides and it was speculated that it is also due to the activity of the APOBEC enzymes. Lastly, deciphering the independent mutational signatures operative in these breast cancer samples provided the means for timing their activity across different cancer subclones [ 86].