gen_graph_indep  R Documentation 
Create an independence graph of genetic differentiation
from genetic data of class genind
Description
The function allows to create genetic graphs from genetic data
by applying the conditional independence principle. Populations whose allelic
frequencies covary significantly once the covariance with the other
populations has been taken into account are linked on the graphs.
Usage
gen_graph_indep(
x,
dist = "basic",
cov = "sq",
pcor = "magwene",
alpha = 0.05,
test = "EED",
adj = "none",
output = "igraph"
)
Arguments
x 
An object of class genind that contains the multilocus
genotype (format 'locus') of the individuals as well as their population
and their geographical coordinates.

dist 
A character string indicating the method used to compute the
multilocus genetic distance between populations
If 'dist = 'basic” (default), then the multilocus genetic distance is
computed using a Euclidean genetic distance formula (Excoffier et al., 1992)
If 'dist = 'weight”, then the multilocus genetic distance is computed
as in Fortuna et al. (2009). It is a Euclidean genetic distance giving more
weight to rare alleles
If 'dist = 'PG”, then the multilocus genetic distance is computed as
in popgraph::popgraph function, following several steps of PCA and SVD
(Dyer et Nason, 2004).
If 'dist = 'PCA”, then the genetic distance is computed following a
PCA of the matrix of allelic frequencies by population. It is a Euclidean
genetic distance between populations in the multidimensional space defined
by all the independent principal components.

cov 
A character string indicating the formula used to compute the
covariance matrix from the distance matrix
If 'cov = 'sq” (default), then the covariance matrix is calculated
from the matrix of squared distances as in Everitt et Hothorn (2011)
If 'cov = 'dist”, then the covariance matrix is calculated from the
matrix of distances as in Dyer et Nason (2004) and popgraph function

pcor 
A character string indicating the way the partial correlation
matrix is computed from the covariance matrix.
If 'pcor = 'magwene”, the steps followed are the same as in
Magwene (2001) and in popgraph::popgraph function. It is the recommended
option as it meets mathematical requirements.
If 'pcor = 'other”, the steps followed are the same as used
by Fortuna et al. (2009). They are not consistent with the approach
of Magwene (2001).

alpha 
A numeric value corresponding to the statistical tolerance
threshold used to test the difference from 0 of the partial correlation
coefficients. By default, 'alpha=0.05'.

test 
A character string indicating the method used to test the
significance of the partial correlation coefficients.
If 'test = 'EED” (default), then the Edge Exclusion Deviance
criterion is used (Whittaker, 2009). Although other methods exist, this is
the most common and thus the only one implemented here.

adj 
A character string indicating the way of adjusting pvalues to
assess the significance of the pvalues
If 'adj = 'none” (default), there is no pvalue adjustment correction
If 'adj = 'holm”, pvalues are adjusted using the sequential
Bonferroni correction (Holm, 1979)
If 'adj = 'bonferroni”, pvalues are adjusted using the classic
Bonferroni correction
If 'adj = 'BH”, pvalues are adjusted using Benjamini et Hochberg
(1995) correction controlling false discovery rate

output 
A character string indicating the matrices included in
the output list.
If 'output = 'all” (default), then D (distance matrix),
C (covariance matrix), Rho (partial correlation matrix),
M (graph incidence matrix) and S (strength matrix) are included
If 'output = 'dist_graph”, then the distance matrix D is returned
only with the values corresponding to the graph edges
If 'output = 'str_graph”, then the strength values matrix S is
returned only with the values corresponding to the graph edges
If 'output = 'inc”, then the binary adjacency matrix M is returned
If 'output = 'igraph”, then a graph of class igraph
is returned

Details
The function allows to vary many parameters such as the genetic
distance used, the formula used to compute the covariance, the statistical
tolerance threshold, the pvalues adjustment, among others.
Value
A list
of objects of class matrix
, an object of
class matrix
or a graph object of class igraph
Author(s)
P. Savary
References
\insertRef
dyer2004populationgraph4lg
\insertRefbenjamini1995controllinggraph4lg
\insertRefbowcock1994highgraph4lg
\insertRefeveritt2011introductiongraph4lg
\insertRefexcoffier1992analysisgraph4lg
\insertReffortuna2009networksgraph4lg
\insertRefholm1979simplegraph4lg
\insertRefmagwene2001newgraph4lg
\insertRefwermuth1977algorithmgraph4lg
\insertRefwhittaker2009graphicalgraph4lg
Examples
data(data_ex_genind)
dist_graph_test < gen_graph_indep(x = data_ex_genind, dist = "basic",
cov = "sq", pcor = "magwene",
alpha = 0.05, test = "EED",
adj = "none", output = "igraph")