Statistical Bioinformatics Webinar: Gong -- Refining Somatic Structural Variant Detection for Precision Oncology

Somatic structural variants (SVs), which are variants that typically impact more than 50
nucleotides, play a significant role in cancer development and evolution, but are
notoriously more difficult to detect than small variants from short-read next-generation
sequencing (NGS) data.  This is due to a combination of challenges attributed to the
purity of tumour samples, tumour heterogeneity, limitations of short-read information
from NGS, and sequence alignment ambiguities.  In spite of active development of SV
detection tools over the past few years, each method has inherent advantages and
limitations.  We aim to evaluate variables impacting our ability to accurately detect
somatic SVs and further facilitate informative decision-making on important impactful
factors.  Using simulation studies, we evaluated single and combinatoric effects of SV
caller, SV types and sizes, variant allele frequency (tumour purity), sequencing depth
of coverage, and variant breakpoint resolution.  Using a generalized additive model
allowed predictions of sensitivity and precision to be made for any combination of
predictors.  The prediction model was implemented in a web-based application, called
Shiny-SoSV, which is freely available at https://hcpcg.shinyapps.io/shiny-sosv.
Shiny-SoSV provides an interactive and visual platform for users to easily compare the
individual and combined impact of different parameters.  It predicts the performance of
a proposed study design on somatic SV detection in silico, prior to the commencement of
experiments.  

For details of previous and upcoming seminars, please visit
https://www.maths.usyd.edu.au/u/SemConf/StatisticalBioinformatics.html