spice: SPICE: Spanning tree based inference of co-expression...
In alorchhota/spice: SPICE: Spanning tree based inference of co-expression networks

Description Usage Arguments Details Value Examples

This function reconstructs a co-expression network from gene expression profiles.

spice(
  expr,
  method = "pearson",
  iter = 100,
  frac.gene = 0.8,
  frac.sample = 0.8,
  n.cores = 1,
  rank.ties = "average",
  filter.mat = NULL,
  weight.method = "qnorm",
  adjust.weight = T,
  adjust.weight.powers = c(seq(1, 10, by = 1), seq(12, 20, by = 2)),
  adjust.weight.rsquared = 0.8,
  adjust.weight.bins = 10,
  adjust.clr = F,
  seed = NULL,
  verbose = F
)

`expr`	matrix or data.frame. Processed gene expression data (gene x sample). The rows must have unique gene names.
`method`	a character string indicating how to compute association between a pair of genes. This must be one of "pearson" (default), "spearman", "mi.empirical", "mi.spearman", and "mi.pearson". See details.
`iter`	numeric. Number of iterations.
`frac.gene`	numeric. Fraction of genes used in each iteration.
`frac.sample`	numeric. Fraction of samples used in each iteration.
`n.cores`	numeric. The number of cores to use.
`rank.ties`	a character string indicating how ties are treated in ranking. Accepted values include `"average", "first", "last", "random", "max", "min", "dense"`. For details, see `ties.method` parameter of `frank`.
`filter.mat`	NULL or a gene x gene matrix indicating which edges to exclude. An edge is excluded if the corresponding entry is `TRUE` or greater than 0. Row and column names of the matrix must be same as `rownames(expr)`.
`weight.method`	a character string indicating how the weights are assigned. It must be either `"qnorm"` (default) or `"inverse.rank"`.
`adjust.weight`	logical. Should the weights be raised to a power to exhibit a scale-free topology.
`adjust.weight.powers`	numeric vector. Powers for which the scale free topology fit indices are to be calculated.
`adjust.weight.rsquared`	numeric. Desired minimum scale free topology fitting index R-squared. If no power achieves `adjust.weight.rsquared`, the power with the maximum R-squared is chosen.
`adjust.weight.bins`	numeric. Number of bins in connectivity histograms used to compute scale-free fit indices.
`adjust.clr`	logical. Should the network weights be adjusted using context likelihood or relatedness? See `clr` for details.
`seed`	integer or NULL. Random number generator seed.
`verbose`	logical or numeric. Print messages if verbose > 0.

The expression matrix expr is expected to be properly normalized, processed and free of batch effects. This function may run slow when there are missing values in expr.

Returns a matrix representing a co-expression network.

n_gene = 10
n_sample = 100
expr = matrix(rnorm(n_gene * n_sample),
              nrow = n_gene,
              ncol = n_sample,
              dimnames = list(sprintf("Gene%s", seq_len(n_gene)),
                              sprintf("Sample%s", seq_len(n_sample))))
spice_net = spice(expr, iter = 10, verbose = TRUE)