run_pcor {dnapath}

Wrapper for partial correlations from corpcor

Description

Conducts co-expression analysis using full partial correlations; these are computed using the shrinkage approach for covariance estimation (Schäfer and Strimmer 2005) from the corpcor package (Schafer et al. 2017). Can be used for the network_inference argument in dnapath.

Usage

run_pcor(x, weights = NULL, ranks = FALSE, verbose = FALSE, ...)

Arguments

x

A n by p matrix of gene expression data (n samples and p genes).

weights

An optional vector of weights. This is used by dnapath() to apply the probabilistic group labels to each observation when estimating the group-specific network.

ranks

If TRUE, the gene expression values will be converted to ranks (across samples) prior to covariance estimation.

verbose

Argument is passed into pcor.shrink.

...

Additional arguments are ignored.

Value

A p by p matrix of association scores.

References

Schäfer J, Strimmer K (2005). “A Shrinkage Approach to Large-Scale Covariance Matrix Estimation and Implications for Functional Genomics.” Statistical Applications in Genetics and Molecular Biology, 4(1), Article 32.

Schafer J, Opgen-Rhein R, Zuber V, Ahdesmaki M, Silva APD, Strimmer. K (2017). corpcor: Efficient Estimation of Covariance and (Partial) Correlation. R package version 1.6.9, https://CRAN.R-project.org/package=corpcor.

See Also

run_aracne, run_bc3net, run_c3net, run_clr, run_corr, run_dwlasso, run_genie3, run_glasso, run_mrnet, and run_silencer

Examples

data(meso)
data(p53_pathways)

# To create a short example, we subset on two pathways from the p53 pathway list,
# and will only run 3 permutations for significance testing.
pathway_list <- p53_pathways[c(8, 13)]
n_perm <- 3

# Use this method to perform differential network analysis.
results <- dnapath(x = meso$gene_expression,
                   pathway_list = pathway_list,
                   group_labels = meso$groups,
                   n_perm = n_perm,
                   network_inference = run_pcor)
summary(results)

# The group-specific association matrices can be extracted using get_networks().
nw_list <- get_networks(results[[1]]) # Get networks for pathway 1.


# nw_list has length 2 and contains the inferred networks for the two groups.
# The gene names are the Entrezgene IDs from the original expression dataset.
# Renaming the genes in the dnapath results to rename those in the networks.
# NOTE: The temporary directory, tempdir(), is used in this example. In practice,
#       this argument can be removed or changed to an existing directory
results <- rename_genes(results, to = "symbol", species = "human",
                        dir_save = tempdir())
nw_list <- get_networks(results[[1]]) # The genes (columns) will have new names.

# (Optional) Plot the network using SeqNet package (based on igraph plotting).
# First rename entrezgene IDs into gene symbols.
SeqNet::plot_network(nw_list[[1]])