07_bayesian_inference

Inferring the age of the wave

We designed a set of scripts which allow you to run the Bayesian analysis that we designed for our data. First of all, the model we assumed as the one that generates our data is the Gamma distribution. Thus, we assume that the normalised branch lengths are distributed following this distribution.

To infer the parameters $(\alpha, \beta)$ linked to this distribution we used a JAGS, a Gibbs sampler which retrieves posterior samples of the parameters and some statistics of the distribution. We designed a series of functions which make this process easier. We also added the Lognormal distribution model to assess the fitting in comparison to the Gamma one.

The module is separated in 3 essential scripts:

src/
├── exec_mcmc.R: the script to execute the analysis using custom inputs
├── mcmc_analysis.R: functions for the MCMC convergence and quality analysis
└── sampling.R: contains the Gibbs sampling functions

W For running the MCMC process which infers the posterior distribution of the Gamma and Lognormal distributions you have to run the following command:

Rscript exec_mcmc.R