set_struct: Gather settings for application of the 'locus' function with...

In hruffieux/locus: Large-Scale Variational Inference for Combined Selection of Covariate and Response Variables in Regression Models

Gather settings for application of the 'locus' function with structured sparse prior.

Description

[FUNCTIONALITY UNDER ACTIVE DEVELOPMENT, PERFORMANCE (CPU TIME) NOT OPTIMIZED]. Posterior probabilities of associations are computed using an empirical covariance estimate of the candidate predictors. This estimate has a block structure (which could reflect linkage disequilibrium patterns when considering genome-wide associations). Such a structure is necessary in large problems for tractability both time- and memory-wise. The posterior probablity of inclusion corresponding to a given block are approximated by a multivariate distribution through a Bernoulli-probit link function.

Usage

set_struct(n, p, pos_st, n_cpus, verbose = TRUE)

Arguments

n	Number of samples.
p	Number of candidate predictors.
pos_st	Vector gathering the predictor block positions (first index of each block). The predictors must be ordered by blocks.
n_cpus	Number of CPUs to be used. Only used if hyper is FALSE, otherwise set it to 1. If large, one should ensure that enough RAM will be available for parallel execution. Set to 1 for serial execution.
verbose	If TRUE, messages are displayed when calling set_struct.

Value

An object of class "struct" preparing the settings for group selection in a form that can be passed to the locus function.

See Also

locus

Examples

seed <- 123; set.seed(seed)

###################
## Simulate data ##
###################

## Example using small problem sizes:
##
n <- 200; p <- 300; p0 <- 100; d <- 50; d0 <- 40

## Candidate predictors (subject to selection)
##
# Here we simulate common genetic variants (but any type of candidate
# predictors can be supplied).
# 0 = homozygous, major allele, 1 = heterozygous, 2 = homozygous, minor allele
#
X_act <- matrix(rbinom(n * p0, size = 2, p = 0.25), nrow = n)
X_inact <- matrix(rbinom(n * (p - p0), size = 2, p = 0.25), nrow = n)

shuff_x_ind <- sample(p)
X <- cbind(X_act, X_inact)[, shuff_x_ind]

bool_x_act <- shuff_x_ind <= p0

pat_act <- beta <- matrix(0, nrow = p0, ncol = d0)
pat_act[sample(p0*d0, floor(p0*d0/5))] <- 1
beta[as.logical(pat_act)] <-  rnorm(sum(pat_act))

## Gaussian responses
##
Y_act <- matrix(rnorm(n * d0, mean = X_act %*% beta, sd = 0.5), nrow = n)
Y_inact <- matrix(rnorm(n * (d - d0), sd = 0.5), nrow = n)
shuff_y_ind <- sample(d)
Y <- cbind(Y_act, Y_inact)[, shuff_y_ind]

########################
## Infer associations ##
########################

n_st <- 100
pos_st <- seq(1, p, by = ceiling(p/n_st))
list_struct <- set_struct(n, p, pos_st, n_cpus = 1)

vb <- locus(Y = Y, X = X, p0_av = p0, link = "identity",
   list_struct = list_struct, user_seed = seed)

hruffieux/locus documentation built on Jan. 10, 2024, 10:07 p.m.