Next-generation DNA sequencing approaches have had a profound impact on our ability to determine the somatic changes that accrue within human cancers. The identification of genes that are recurrently mutated within cancer subtypes provides salient clues into the oncogenic mechanisms occurring within these cells and can be suggestive of novel therapeutic approaches. For example, the FOXL2 gene in granulosa cell ovarian cancer and the EZH2 gene in lymphoma both have a highly recurrent mutation within a specific codon that was identified by next generation sequencing. At the Genome Sciences Centre, Vancouver, we are developing computational pipelines that are able to detect recurrent mutations, using both reference alignment and de-novo assembly approaches. De-novo assembly approaches having the advantage of precisely reconstructing sequences across rearrangement breakpoints, which often incorporate non-templated sequences. Importantly the transcriptome, which represents the “read-out” of the genome, can also harbor specific changes that can contribute to the oncogenic status of a cell. Changes such as novel transcript isoforms, RNA edits, duplicated or novel exons and the in-frame expression of gene fusions are difficult to predict from genomic sequence alone. Therefore we have also utilized the Trans-Abyss transcriptome assembly approach to specifically attempt to reconstruct and identify such features in cancer transcriptomes. We have also established a database schema and API which allows the storage and retrieval of variants determined by next generation sequence data. This allows the rapid determination of the oncogenic relevance of any detected sequence variants within other tumour types and the frequency of such variants in the normal population. Correlating the observed changes in tumours with appropriate therapeutic intervention or using the mutational status to infer novel therapeutic targets remain key goals and challenges in our work.
Back to Workshop I: Next-generation Sequencing Technology and Algorithms for Primary Data Analysis