Skip to contents

Draw samples from posterior distributions of fitted models

Source: R/get_posterior.R
getPosteriorSamples.Rd

This is an optional step in an analysis using limorhyde2, and is useful for quantifying uncertainty in posterior estimates of fitted curves and rhythmic statistics. The function calls mashr::mash_compute_posterior_matrices().

Usage

getPosteriorSamples(fit, nPosteriorSamples = 200L, overwrite = FALSE)

Arguments

fit

A `limorhyde2' object containing posterior fits.

nPosteriorSamples

Number of samples to draw from each posterior distribution.

overwrite

Logical indicating whether to recompute posterior samples if they already exist.

Value

A limorhyde2 object containing everything in fit with added or updated element:

  • mashPosteriorSamples: a three-dimensional array of coefficients, with dim 1 corresponding to features, dim 2 to model terms, and dim 3 to posterior samples.

See also

getPosteriorFit(), getRhythmStats(), getExpectedMeas(), getStatsIntervals()

Examples

library('data.table')

y = GSE54650$y
metadata = GSE54650$metadata

fit = getModelFit(y, metadata)
fit = getPosteriorFit(fit)
fit = getPosteriorSamples(fit, nPosteriorSamples = 10L)

rhyStatsSamps = getRhythmStats(
  fit, features = c('13170', '13869'), fitType = 'posterior_samples')
rhyStatsInts = getStatsIntervals(rhyStatsSamps)