bpr_bayes: Gibbs sampling approach for the BPR model

In andreaskapou/BPRMeth-devel: Model higher-order methylation profiles

Description

The function bpr_bayes computes the posterior of the BPR model using auxiliary variable approach. Since we cannot compute the posterior analytically, a Gibbs sampling scheme is used.

Usage

bpr_bayes(x, ...)

## S3 method for class 'list'
bpr_bayes(x, w_mle = NULL, basis = NULL,
  fit_feature = NULL, cpg_dens_feat = FALSE, w_0_mean = NULL,
  w_0_cov = NULL, gibbs_nsim = 20, gibbs_burn_in = 10,
  keep_gibbs_draws = FALSE, is_parallel = TRUE, no_cores = NULL, ...)

## S3 method for class 'matrix'
bpr_bayes(x, w_mle = NULL, basis = NULL,
  fit_feature = NULL, cpg_dens_feat = FALSE, w_0_mean = NULL,
  w_0_cov = NULL, gibbs_nsim = 20, gibbs_burn_in = 10,
  keep_gibbs_draws = FALSE, ...)

Arguments

x	The input object, either a matrix or a list.
...	Additional parameters.
w_mle	A vector of parameters (i.e. coefficients of the basis functions) containing the MLE estimates.
basis	A 'basis' object. E.g. see create_rbf_object.
fit_feature	Return additional feature on how well the profile fits the methylation data. Either NULL for ignoring this feature or one of the following: 1) "RMSE" for returning the fit of the profile using the RMSE as measure of error or 2) "NLL" for returning the fit of the profile using the Negative Log Likelihood as measure of error.
cpg_dens_feat	Logical, whether to return an additional feature for the CpG density across the promoter region.
w_0_mean	The prior mean hyperparameter for w
w_0_cov	The prior covariance hyperparameter for w
gibbs_nsim	Optional argument giving the number of simulations of the Gibbs sampler.
gibbs_burn_in	Optional argument giving the burn in period of the Gibbs sampler.
keep_gibbs_draws	Logical indicating if we should keep the whole MCMC chain for further analysis.
is_parallel	Logical, indicating if code should be run in parallel.
no_cores	Number of cores to be used, default is max_no_cores - 2.

Value

Depending on the input object x:

• If x is a list: An object containing the following elements:

  • W_opt: An Nx(M+1) matrix with the optimized parameter values. Each row of the matrix corresponds to each element of the list x. The columns are of the same length as the parameter vector w (i.e. number of basis functions).

  • Mus: An N x M matrix with the RBF centers if basis object is create_rbf_object, otherwise NULL.

  • basis: The basis object.

  • w: The initial values of the parameters w.

• If x is a matrix: An object containing the following elements:

  • w_opt: Optimized values for the coefficient vector w. The length of the result is the same as the length of the vector w.

  • basis: The basis object.

Author(s)

C.A.Kapourani [e-mail]

See Also

create_basis, eval_functions

Examples

data <- meth_data
out_opt <- bpr_bayes(x = data, is_parallel = FALSE)

ex_data <- meth_data
basis <- create_rbf_object(M=3)
out_opt <- bpr_bayes(x = ex_data, is_parallel = FALSE, basis = basis)

basis <- create_rbf_object(M=2)
w <- c(0.1, 0.1, 0.1)
w_0_mean <- rep(0, length(w))
w_0_cov <- diag(10, length(w))
data <- meth_data[[1]]
out_opt <- bpr_bayes(x = data, w_mle = w, w_0_mean = w_0_mean,
                     w_0_cov = w_0_cov, basis = basis)

basis <- create_rbf_object(M=0)
w <- c(0.1)
w_0_mean <- rep(0, length(w))
w_0_cov <- diag(10, length(w))
data <- meth_data[[1]]
out_opt <- bpr_bayes(x = data, w_mle = w, w_0_mean = w_0_mean,
                     w_0_cov = w_0_cov, basis = basis)

andreaskapou/BPRMeth-devel documentation built on May 12, 2019, 3:32 a.m.