Perform Full FASH Analysis

This function performs the full FASH pipeline, including data setup, likelihood computation, empirical Bayes estimation, and outputs a structured fash object.

fash(
  Y,
  smooth_var,
  offset = 0,
  S = NULL,
  Omega = NULL,
  data_list = NULL,
  grid = seq(0, 1, length.out = 25),
  likelihood = c("gaussian", "poisson"),
  num_basis = 30,
  betaprec = 1e-06,
  order = 1,
  pred_step = 1,
  penalty = 1,
  num_cores = 1,
  verbose = TRUE
)

Arguments

Y

Either a numeric matrix of response variables or a character string specifying the column name in data_list for response variables.

smooth_var

A numeric matrix, vector, or a character string specifying the column name in data_list for smoothing variables.

offset

A numeric matrix, vector, scalar, or a character string specifying the column name in data_list for offset variables.

S

A numeric matrix, vector, scalar, or list representing the standard errors of Y. Or a character string specifying the column name in data_list for SD.

Omega

Either a list of precision matrices (one for each dataset) or a single precision matrix (shared across all datasets).

data_list

A list of data frames, where each data frame corresponds to a single dataset.

grid

A numeric vector representing the grid of PSD (Predictive Standard Deviation) values.

likelihood

A character string specifying the likelihood function to use. Options are `gaussian` and `poisson`.

num_basis

An integer specifying the number of O-Spline basis functions.

betaprec

A numeric value representing the precision of the fixed effects coefficients.

order

A positive integer specifying the order of the Integrated Wiener Process (IWP) prior.

pred_step

A numeric value specifying the prediction step size.

penalty

A numeric value representing the lambda value for the Dirichlet prior.

num_cores

An integer specifying the number of cores to use for parallel processing.

verbose

A logical value. If TRUE, shows progress messages and timing for each step.

Value

A fash object containing:

prior_weights

Estimated prior weights for PSD values.

posterior_weights

Posterior weight matrix of PSD values.

psd_grid

PSD grid values.

lfdr

Local false discovery rate for each dataset.

settings

A list of settings used in the FASH pipeline.

fash_data

A structured list of data components.

L_matrix

Likelihood matrix used in the FASH pipeline.

eb_result

Empirical Bayes estimation results.

Examples

set.seed(1)
data_list <- list(
  data.frame(y = rpois(5, lambda = 5), x = 1:5, offset = 0),
  data.frame(y = rpois(5, lambda = 5), x = 1:5, offset = 0)
)
grid <- seq(0, 2, length.out = 10)
result <- fash(data_list = data_list, Y = "y", smooth_var = "x", offset = "offset", grid = grid, likelihood = "poisson", verbose = TRUE)
#> Starting data setup...
#> Completed data setup in 0.00 seconds.
#> Starting likelihood computation...
#> 
  |                                                                            
  |                                                                      |   0%
  |                                                                            
  |===================================                                   |  50%
  |                                                                            
  |======================================================================| 100%
#> Completed likelihood computation in 0.07 seconds.
#> Starting empirical Bayes estimation...
#> Completed empirical Bayes estimation in 0.00 seconds.
#> fash object created successfully.
print(result)
#> Fitted fash Object
#> -------------------
#> Number of datasets: 2
#> Likelihood: poisson
#> Number of PSD grid values: 10 (initial), 1 (non-trivial)
#> Order of Integrated Wiener Process (IWP): 1