Earlier this year I was fortunate enough to work with EcoHealth Alliance on a major project focused on two WHO priority zoonoses, Rift Valley fever virus and Crimean-Congo hemorrhagic fever virus. Over a 10 year period, research teams sampled over 800 people and livestock, sampled over 300 rodents, and conducted ecological site characterizations at 150 sites in each Tanzania and South Africa.
To integrate data from disparate datasets we developed a novel R pipeline to standardize data cleaning, quality control and integration following reproducible science principals. This was one of the first projects of this scale that I have seen using so many tools that align to current industry best practice:
- Version controlled code in Github with strict code review
- Full CI/CD automation with Github Actions
{targets}as a pipeline tool
- R package to store and document all functions
{renv}for package dependency management
- Proper encrypted environment files for all credentials
A common deterrent I hear from beginners around using tools to aid reproducibility in R code often center around the barriers to entry. They are seen as adding complexity and overhead into the process. Which they do (on small projects), but as projects become more and more complex, the benefits quickly outweigh the costs.
An overview of this project was given as a poster at the recent 8th World One Health Congress in Cape Town
Want to read more? Sign up to my mailing list here for weekly emails and no spam.
Looking for a data science consultant? Feel free to get in touch at wavedatalabs.com.au