2019-nCoV-SARS/COVID-19 pandemic response
Our team is providing technical and software support to the ongoing 2019-nCoV-SARS/COVID-19 pandemic. Much of this builds on tools that we have developed over the previous 5 years, with some ongoing development of these tools to track the research needs:
- orderly and OrderlyWeb, our reproducible reporting framework developed for the Vaccine Impact Modelling Consortium, is being used to coordinate research results amongst our (now distributed) team of researchers in a reproducible and traceable way. During the response we have developed extensions such as sharepoint support and an Rstudio addin, as well as a lot of new features
- odin is being used to implement both stochastic and deterministic compartmental models, and we have developed odin.js which can compile odin models to JavaScript during the pandemic to deploy models to the web. For stochastic models we have started developing a new simulation engine (dust) and tools for inference (mcstate)
- cyphr is being used to encrypt data at rest
We are working directly with researchers as they develop models and analyses, supporting
- CovidSim (an individual-based simulation requiring significant HPC capacity to run)
- sircovid an odin model for the epidemic through the UK hospital system, fit using particle filter MCMC
- squire an odin model with both a stochastic version used to fit to data (with nightly reports fitting to 110 countries) and a deterministic version that is available on the web (the web app was created by Bio Nano consulting, and the embedded model runs purely client side, compiled with odin.js). Featured on Our World in Data
- Tooling and strategies for processing multiple daily streams of incoming unstandardised data.