dependentsimr (version 1.0.0.0)

get_random_structure: Compute structure of dependency from a given data

Description

Compute structure of dependency from a given data

Usage

get_random_structure(datasets, method, rank = 2, types = "normal")

Value

A random structure element suitable for use with draw_from_multivariate_corr().

Arguments

datasets

A list of data matrices that we will mimic. All datasets have samples in the columns and features (e.g., genes, proteins) in the rows. All datasets must have the same samples in corresponding columns.

method

One of 'pca', 'spiked Wishart', or 'corp.cor', for the method of determining a dependency structure

rank

Number of PCA components to approximate the dependence structure of. Only used if method = 'pca'.

types

The marginal distribution types ('normal', 'poisson', 'DESeq2', or 'empirical'), as a list with entries corresponding to datasets. If just a single value is provided, then it is used for all datasets.

Examples

# Simple data to mimic
true_means <- c(0, 1, 10, 1000)
data <- rpois(4*12, true_means) |> matrix(4, 12)

# Simulate draws mimicking that data
rs <- get_random_structure(list(data=data), method="pca", rank=2, type="poisson")
draws <- draw_from_multivariate_corr(rs, n_samples=30)