Estimation of the serial interval of novel coronavirus (COVID-19) infections
Supporting materials for Nishiura H, Linton NM, Akhmetzhanov AR 2020 "Serial interval of novel coronavirus (COVID-19) infections" Submitted
Our repository consists of the following parts:
1. Inference of the serial interval using only infector-infectee pairs identified from the literature with high certainty
- A1. Stan simulations.ipynb Code to run MCMC simulations in cmdStan
- A2. Processing the traces.ipynb Python script to analyse posterior distirubions generated by Stan
2. Inference of the serial interval using only infector-infectee pairs identified as certain or probable
- B1. Stan simulations including probable cases.ipynb Code to run MCMC simulations in cmdStan
- B2. Processing the traces for certain and probable cases.ipynb Python script to analyse posterior distirubions generated by Stan
3. Generating Figure 1
4. Additional details
- The folder data contains the supplementary table used for our analysis.
- The folder results contains all constructed traceplots, as well as individual traces for the parameters of the distributions. As such, param1 & param2 in csv or pickle files are respectively the meanlog and sdlog for the lognormal distribution, the shape and inverse scale for the gamma distribution, and the shape and scale for the Weibull distribution. We followed the notation according to Stan manual (see the pages on lognormal, gamma, and Weibull distributions). The
scipy.statsnotation for each distribution follows the code:
ss.weibull_min(param1,scale=param2), respectively. Here
npare shortcuts for
numpylibraries in Python.
Thank you for your interest to our work!
Few words of caution: We would like to note that our code is not supposed to work out of box, because the links used in the notebooks were user-specific, and our main intent was to show the relevance of the methods used in our paper.