R/Inference.R

In snaketron/MultilevelGwas: Genome wide association studies of multiple traits with Bayesian multilvel models

# Description:
# Bayesian inference
runInference <- function(gt.data,
                         mcmc.chains,
                         mcmc.steps,
                         mcmc.warmup,
                         mcmc.cores,
                         stan.model,
                         adapt.delta,
                         max.treedepth,
                         comparison,
                         write.samples) {


  data.list <- gt.data

  # get appropriate parameters to monitor
  pars <- getStanModelPars(model.name = stan.model@model_name)


  # sample file
  sample.file <- NULL
  if(write.samples == TRUE) {
    sample.file <- paste(round(x = runif(n=1,min=0,max=10^6), digits=0),
                         "sampling", stan.model@model_name, sep = '_')
  }


  # run
  glm <- rstan::sampling(object = stan.model,
                         data = data.list,
                         iter = mcmc.steps,
                         warmup = mcmc.warmup,
                         chains = mcmc.chains,
                         cores = mcmc.cores,
                         control = list(adapt_delta = adapt.delta,
                                        max_treedepth = max.treedepth),
                         pars = pars,
                         include = TRUE,
                         sample_file = sample.file,
                         refresh = 100,
                         algorithm = "NUTS")

  return (list(glm = glm))
}