Fitted mfsusie objects under five preprocessing configurations

Five fits on the same simulated dataset (n = 40, p = 25, M = 2, T_m = 32, signal SNP at index 7), one per configuration: default (Wakefield Normal marginal), small-sample Student-t correction, low-count column filtering, wavelet-domain quantile normalization, and all three corrections combined. The simulated dataset reproduces the shape of a small-cohort chromatin-accessibility QTL panel (sparse coverage in some cell types, heavy-tailed wavelet coefficients in others). The underlying genotype and read-count data are not redistributable; this object ships fitted-model summaries only, no individual-level inputs.

Usage

practical_fits

Format

A list with components

n, p, M, T_m, signal_snp: simulation metadata.
cell_types: character vector of length M naming the two simulated cell types.
default: mfsusie fit with all defaults (Wakefield Normal marginal Bayes factor).
johnson: mfsusie fit with small_sample_correction = TRUE (Johnson-t marginal).
lowcount: mfsusie fit with wavelet_magnitude_cutoff = 0.02.
qn: mfsusie fit with wavelet_qnorm = TRUE.
combined: mfsusie fit with all three corrections enabled.

Each fit is stripped of Y_grid, X_eff, tracking, and any raw X/Y matrices to keep the package size small and avoid shipping individual-level data.

Source

Simulated. See data-raw/make_practical_dataset.R.

Examples

# \donttest{
data(practical_fits)
practical_fits$default$pip[practical_fits$signal_snp]
#> [1] 0.9999609
# }