Profile likelihood ratio test with PACS

Profile likelihood ratio test with PACS

Usage

pacs_test_sparse(
  covariate_meta.data,
  formula_full,
  formula_null,
  pic_matrix,
  n_peaks_per_round = NULL,
  T_proportion_cutoff = 0.2,
  cap_rates,
  par_initial_null = NULL,
  par_initial_full = NULL,
  n_cores = 1,
  verbose = TRUE
)

Arguments

covariate_meta.data

A data.frame with columns representing the covariates and rows representing cells

formula_full

A formula object representing the full model. For example, ~ cell_type + batch

formula_null

A formula object representing the null model. For example, ~ batch

pic_matrix

The input region-by-cell PIC matrix

n_peaks_per_round

The number of peaks tested in each round. Default = NULL, which will be determined by the maximum allowable size of matrix in R.

T_proportion_cutoff

To determine the maximum value of T, we set a criteria based on the proportion of reads that are >= T. This parameter set this proportion. Default = 0.2

cap_rates

A vector of capturing probability for each cell

par_initial_null

Initialized values of estimated parameters for the null model, we do not need to specify unless there are reasons to do so. Default = NULL

par_initial_full

Initialized values of estimated parameters for the null model, we do not need to specify unless there are reasons to do so. Default = NULL

n_cores

number of cores for multi-core computation

verbose

Whether to print progress, default = TRUE

Value

A list of two elements, pacs_converged is a vector of length 2*n_peaks representing the convergence status of the peak in the null and full model and pacs_p_val is a vector of length n_peaks representing the p values for each peak.