bootstrap_rho function

Perform a parametric bootstrapping correction on an estimated rho vector

Takes an estimate of rho, and a two-column format genetic data frame containing both reference and mixture data. Returns a new rho corrected by parametric bootstrapping

bootstrap_rho(
  rho_est,
  pi_est,
  D,
  gen_start_col,
  niter = 100,
  reps = 2000,
  burn_in = 100,
  pi_prior = NA,
  pi_prior_sum = 1
)

Arguments

• rho_est: the rho value previously estimated from MCMC GSI from the provided reference and mixture data

• pi_est: the pi value previously estimated from MCMC GSI from the provided reference and mixture data

• D: a two-column genetic dataframe containing the reference and mixture data from which rho_est was computed; with "repunit", "collection", and "indiv" columns

• gen_start_col: the first column of genetic data in D. All columns after gen_start_col must be genetic data

• pi_prior: The prior to be added to the collection allocations, in order to generate pseudo-count Dirichlet parameters for the simulation of a new pi vector. Non-default values should be a vector of length equal to the number of populations in the reference dataset. Default value of NA leads to the calculation of a symmetrical prior based on pi_prior_sum.

• pi_prior_sum: total weight on default symmetrical prior for pi.

  In parametric bootstrapping, niter new mixture datasets are simulated by individual from the reference with reporting unit proportions rho_est, and the mean of their MCMC GSI outputs is used to calculate an average bias. This bias is subtracted from rho_est to give the output. The number of individuals in each simulated bootstrap dataset is equal to the number of "mixture" individuals in D.

Returns

bootstrap_rho returns a new rho value, corrected by parametric bootstrapping.