The amount of biological data being generated has exploded in the past 5 years, primarily due to the decreasing costs of high throughput sequencing. In particular tens of thousands of cancer genomes have been sequenced. Interpreting this data in a meaningful way however, is challenging. We hypothesized that the patterns of genetic changes in cancers genomes that we observe are a consequence of how cancers grow. To test this we developed a stochastic simulation of tumor growth, which we could fit to cancer genomic data using the Approximate Bayesian Computation (ABC) framework. Due to Julia’s unique capabilities we implemented our stochastic simulations as well as the statistical inference in the language. Approximate Bayesian methods require potentially millions of simulations with different parameters to accurately fit to data, Julia’s speed and ease of use makes it an attractive option for such problems. Marc will also touch on why Julia holds huge promise for other “big data” problems and how he’s been using Julia as an open research tool with other projects such as Jupyter notebooks.

YOU MAY ALSO LIKE:

• droidcon London 2017 (in London on 26th - 27th October 2017)
• Brian Sletten's Data Science with R Workshop (in London on 8th - 10th November 2017)