get_fit.cor: Calculating best fit of a power low on correlation matrix...

In Kumquatum/GWENA: Pipeline for augmented co-expression analysis

Calculating best fit of a power low on correlation matrix computed on expression data

Description

Adjust a correlation matrix depending of the type of network, then try to parameter a power law for best fit

Usage

get_fit.cor(
  cor_mat,
  fit_cut_off = 0.9,
  network_type = c("unsigned", "signed", "signed hybrid"),
  block_size = NULL,
  ...
)

Arguments

cor_mat	matrix or data.frame of genes correlation.
fit_cut_off	float, cut off by which R^2 (coefficient of determination) will be thresholded. Must be in ]0;1[.
network_type	string giving type of network to be used. Either "unsigned", "signed", "signed hybrid". See details.
block_size	integer giving size of blocks by which operations can be proceed. Helping if working with low capacity computers. If null, will be estimated.
...	any other parameter compatible with pickSoftThreshold.fromSimilarity

Details

network_type indicate which transformation will be applied on the correlation matrix to return the similarity score.

signed

will modify the range [-1;1] to [0.5;1.5] (because of log10 beeing used for scale free index computation)

unsigned

will return absolute value (moving from [-1;1] to [0;1])

signed hybrid

will replace all negative values by 0 (moving from [-1;1] to [0;1])

Value

A list containing power of the law for best fit, fit table, and metadata about the arguments used.

Examples

get_fit.cor(cor_mat = cor(kuehne_expr[, seq_len(100)]))

Kumquatum/GWENA documentation built on July 7, 2023, 3:41 p.m.