R/PRE_FATE.speciesClustering_step2.R
In MayaGueguen/RFate: Vegetation modelling with the FATE model
Defines functions
PRE_FATE.speciesClustering_step2
Documented in
PRE_FATE.speciesClustering_step2
### HEADER #####################################################################
##' @title Choose clusters and select determinant species
##'
##' @name PRE_FATE.speciesClustering_step2
##'
##' @author Maya Guéguen
##'
##' @description This script is designed to obtain functional groups by : 1)
##' selecting the number of clusters to be kept from an object obtained with
##' the \code{\link{PRE_FATE.speciesClustering_step1}} function ; 2) refining
##' these groups by identifying determinant species in each of them.
##'
##' @param clust.dendrograms a dendrogram, or a \code{list} of dendrograms (one
##' for each \code{GROUP} value). \cr Such an object can be obtained
##' with the \code{\link{PRE_FATE.speciesClustering_step1}} function.
##' @param no.clusters an \code{integer}, or a \code{vector} of \code{integer}
##' (one for each \code{GROUP} value), with the number of clusters to be kept
##' @param mat.species.DIST a \code{dist} object, or a \code{list} of
##' \code{dist} objects (one for each \code{GROUP} value), corresponding to the
##' distance between each pair of species. \cr Such an object can be obtained
##' with the \code{\link{PRE_FATE.speciesDistance}} function.
##'
##'
##' @details
##'
##' This function allows to obtain a classification of \strong{dominant}
##' species into Plant Functional Groups (PFG), and the \strong{determinant}
##' species based on these PFG. \cr \cr
##'
##' \strong{What is the difference between \code{dominant} and
##' \code{determinant} species ?}
##'
##' \itemize{
##' \item{\strong{Dominant} species are species representative of an
##' environment or a studied area, in terms of number of releves or
##' abundance values. They can be found with the
##' \code{\link{PRE_FATE.selectDominant}} function of this package. These
##' dominant species are used to build PFG with the
##' \code{\link{PRE_FATE.speciesClustering_step1}} function.
##' }
##' \item{Once PFG are built, \strong{determinant} species are defined as
##' refined subsets of dominant species within each PFG. \cr The process is
##' detailed below :
##' \itemize{
##' \item each dominant species is assigned to a PFG
##' \item within each PFG :
##' \itemize{
##' \item for each species, compute its mean distance to the other
##' species within the PFG (\code{sp.mean.dist})
##' \item calculate the mean value of all these mean distances
##' (\code{allSp.mean})
##' \item calculate the deviation values around this mean value
##' (\code{allSp.min} and \code{allSp.max})
##' \item determinant species are the ones that are included between
##' those deviation values
##' }
##' }
##' }
##' }
##'
##' @return A \code{list} containing one \code{vector}, one \code{data.frame}
##' with the following columns, and two \code{ggplot2} objects :
##'
##' \describe{
##' \item{determ.sp}{the names of all determinant species}
##' \item{determ.all}{(\emph{determinant and non-determinant species})
##' \describe{
##' \item{\code{PFG}}{ID of the plant functional group
##' (\code{GROUP} + \code{ID.cluster})}
##' \item{\code{GROUP}}{name of data subset}
##' \item{\code{ID.cluster}}{cluster number}
##' \item{\code{species}}{name of species}
##' \item{\code{ID.species}}{species number in each PFG}
##' \item{\code{sp.mean.dist}}{species mean distance to other species of
##' the same PFG}
##' \item{\code{allSp.mean}}{\eqn{mean(\text{sp.mean.dist})} within the PFG}
##' \item{\code{allSp.min}}{\eqn{mean(\text{sp.mean.dist}) - 1.64 *
##' sd(\text{sp.mean.dist})} within the PFG}
##' \item{\code{allSp.max}}{\eqn{mean(\text{sp.mean.dist}) + 1.64 *
##' sd(\text{sp.mean.dist})} within the PFG}
##' \item{\code{DETERMINANT}}{\code{TRUE} if determinant species, \code{FALSE}
##' otherwise}
##' }
##' }
##' \item{plot.distance}{\code{ggplot2} object, representing the distribution
##' of mean distances between species for each functional group}
##' \item{plot.PCO}{\code{list} of \code{ggplot2} objects, representing the
##' PFG within the functional space \cr \cr}
##' }
##'
##' One \file{PRE_FATE_CLUSTERING_STEP_2_distantSpecies_PCO.pdf} file is created
##' containing two types of graphics :
##' \describe{
##' \item{distantSpecies}{to visualize in each PFG the distribution of mean
##' distance of each species to other species, and non-determinant species
##' which are outside the distribution}
##' \item{PCO}{to visualize in each PFG the distribution of species, with and
##' without non-determinant species}
##' }
##'
##' @keywords hierarchical clustering, Principal Component Ordination
##'
##' @seealso \code{\link[stats]{cutree}},
##' \code{\link[ade4]{quasieuclid}},
##' \code{\link[ade4]{dudi.pco}},
##' \code{\link{PRE_FATE.speciesDistance}},
##' \code{\link{PRE_FATE.speciesClustering_step1}},
##' \code{\link{PRE_FATE.speciesClustering_step3}}
##'
##' @examples
##'
##' ## Load example data
##' data(DATASET_Bauges_PFG)
##'
##' ## Species dissimilarity distance (niche overlap + traits distance)
##' tab.dist = DATASET_Bauges_PFG$dom.dist_total
##' str(tab.dist)
##' as.matrix(tab.dist[[1]])[1:5, 1:5]
##'
##'
##' ## Build dendrograms -------------------------------------------------------------------------
##' sp.CLUST = PRE_FATE.speciesClustering_step1(mat.species.DIST = tab.dist)
##' names(sp.CLUST)
##'
##'
##' ## Number of clusters per group
##' plot(sp.CLUST$plot.clustNo)
##' no.clusters = c(5, 5, 9)
##'
##' ## Find determinant species ------------------------------------------------------------------
##' sp.DETERM = PRE_FATE.speciesClustering_step2(clust.dendrograms = sp.CLUST$clust.dendrograms
##' , no.clusters = no.clusters
##' , mat.species.DIST = tab.dist)
##' names(sp.DETERM)
##'
##' str(sp.DETERM$determ.sp)
##' str(sp.DETERM$determ.all)
##'
##' ## Species names
##' sp.NAMES = DATASET_Bauges_PFG$sp.names
##' sp.NAMES$species = paste0("X", sp.NAMES$species)
##' determ = merge(sp.DETERM$determ.all, sp.NAMES, by = "species", all.x = TRUE)
##' str(determ)
##'
##' plot(sp.DETERM$plot.distance)
##' plot(sp.DETERM$plot.PCO$Chamaephyte)
##' plot(sp.DETERM$plot.PCO$Herbaceous)
##' plot(sp.DETERM$plot.PCO$Phanerophyte)
##'
##'
##' @export
##'
##' @importFrom grDevices colorRampPalette pdf dev.off
##' @importFrom stats as.dist cutree sd
##' @importFrom graphics plot
##'
##' @importFrom ade4 quasieuclid dudi.pco inertia.dudi
##'
##' @importFrom ggplot2 ggplot ggsave aes_string
##' geom_point geom_hline geom_vline geom_errorbar geom_path
##' scale_color_discrete scale_color_manual scale_shape_manual scale_size_manual
##' facet_grid labs theme element_text element_blank
##' @importFrom ggthemes theme_fivethirtyeight
##' @importFrom ggrepel geom_label_repel
##'
## END OF HEADER ###############################################################
PRE_FATE.speciesClustering_step2 = function(clust.dendrograms
, no.clusters
, mat.species.DIST
){
#############################################################################
## CHECK parameter clust.dendrograms
if (.testParam_notInClass(clust.dendrograms, c("list","hclust")))
{
stop("No data given!\n (missing `clust.dendrograms` information which must be of class `hclust` or a list `hclust` objects)")
} else
{
if (class(clust.dendrograms) == "list" &&
length(which(sapply(clust.dendrograms, class) == "hclust")) < length(clust.dendrograms))
{
stop("Wrong type of data!\n each element of `clust.dendrograms` must be of class `hclust`")
}
if (class(clust.dendrograms) == "hclust")
{
clust.dendrograms = list(GROUP1 = clust.dendrograms)
}
if(!is.null(names(clust.dendrograms)))
{
group_names = names(clust.dendrograms)
} else
{
group_names = paste0("GROUP", 1:length(clust.dendrograms))
}
}
## CHECK parameter no.clusters
if (.testParam_notNum(no.clusters))
{
stop("No data given!\n (missing `no.clusters` information)")
} else
{
if (length(no.clusters) != length(clust.dendrograms))
{
stop("Wrong type of data!\n `no.clusters` must have the same length than `clust.dendrograms`")
}
}
## CHECK parameter mat.species.DIST
if (.testParam_notInClass(mat.species.DIST, c("list","dist")))
{
stop("No data given!\n (missing `mat.species.DIST` information which must be a dist object, or a list of dist objects)")
} else
{
if (class(mat.species.DIST) == "list" &&
length(mat.species.DIST) != length(clust.dendrograms))
{
stop("Wrong type of data!\n `mat.species.DIST` must have the same length than `clust.dendrograms`")
}
if (class(mat.species.DIST) == "dist")
{
mat.species.DIST = list(GROUP1 = mat.species.DIST)
}
}
cat("\n\n #------------------------------------------------------------#")
cat("\n # PRE_FATE.speciesClustering_step2 : DETERMINANT SPECIES")
cat("\n #------------------------------------------------------------# \n")
#############################################################################
## DEFINITION OF CLUSTERED GROUPS
#############################################################################
### CUT dendrogramS (or trees) RESULTING FROM hclust INTO SEVERAL GROUPS (no = k)
## Number of groups for each group has been chosen
## according to the previous plot (CLUSTERING_STEP_1B)
clust.groups = lapply(1:length(group_names), function(x)
cutree(clust.dendrograms[[x]], k = no.clusters[x]))
clust.groups = lapply(1:length(group_names), function(x) {
tmp.names = names(clust.groups[[x]])
tmp = paste0(group_names[x], ".", clust.groups[[x]])
names(tmp) = tmp.names
return(tmp)
})
clust.groups = unlist(clust.groups)
PFG_names = sort(unique(clust.groups))
#############################################################################
## IDENTIFY DETERMINANT SPECIES
#############################################################################
## DETERMINANT SPECIES are the ones whose mean distance to other species
## is in the distribution of mean distances of every species to other species
determ = foreach(pfg = PFG_names) %do%
{
## get the SPECIES names
sp = names(clust.groups)[which(clust.groups == pfg)]
if (length(sp) > 1)
{
## get the GROUP information
group = strsplit(pfg, "[.]")[[1]][1]
ID.cluster = strsplit(pfg, "[.]")[[1]][2]
## get the DISTANCE information
mat = as.matrix(mat.species.DIST[[group]])[sp, sp]
## compute for each species its mean distance to other species
sp.mean.dist = apply(mat, 1, mean, na.rm = T)
return(data.frame(PFG = pfg
, GROUP = group
, ID.cluster
, species = sp
, ID = 1:length(sp)
, sp.mean.dist
, allSp.mean = mean(sp.mean.dist)
, allSp.min = mean(sp.mean.dist) - 1.64 * sd(sp.mean.dist)
, allSp.max = mean(sp.mean.dist) + 1.64 * sd(sp.mean.dist)
, stringsAsFactors = FALSE))
} else if (length(sp) == 1)
{
return(data.frame(PFG = pfg
, GROUP = group
, ID.cluster
, species = sp
, ID = 1
, sp.mean.dist = NA
, allSp.mean = NA
, allSp.min = NA
, allSp.max = NA
, stringsAsFactors = FALSE))
}
}
determ = do.call(rbind, determ)
rownames(determ) = determ$species
if (length(determ) == 0 || is.null(determ))
{
stop("No determinant species have been selected. Please check your data")
}
determ$DETERMINANT = TRUE
determ$DETERMINANT[which(determ$sp.mean.dist > determ$allSp.max)] = FALSE
determ$DETERMINANT[which(determ$sp.mean.dist < determ$allSp.min)] = FALSE
determ$DETERMINANT = factor(determ$DETERMINANT, c(TRUE, FALSE))
## SAVE DETERM
## CAT INFO ABOUT HOW MANY SPECIES ARE REMOVED
## CAT INFO ABOUT HOW MANY SPECIES in each PFG
#############################################################################
## GRAPHICAL REPRESENTATIONS
#############################################################################
## GRAPHICAL REPRESENTATION 1 -----------------------------------------------
colRamp = colorRampPalette(c('#8e0152','#c51b7d','#de77ae','#7fbc41','#4d9221','#276419'))
colLev = levels(interaction(determ$DETERMINANT, determ$GROUP))
pp3 = ggplot(determ, aes_string(x = "PFG", y = "sp.mean.dist"
, color = interaction(determ$DETERMINANT, determ$GROUP)
, shape = "DETERMINANT")) +
scale_color_manual(guide = F, values = colRamp(length(colLev))) +
scale_shape_manual(guide = F, values = c("0" = 20, "1" = 8)) +
geom_errorbar(aes_string(ymin = "allSp.min", ymax = "allSp.max")
, color = "darkblue") +
geom_point(position = "jitter") +
geom_point(aes_string(y = "allSp.mean")
, pch = 18
, lwd = 5
, color = "darkblue") +
facet_grid("~ GROUP", scales = "free_x") +
labs(x = "", y = "Mean distance to other species"
, title = "STEP C : Removal of distant species"
, subtitle = paste0("Only species whose mean distance to other species "
, "is included in the distribution\n"
, "of all PFG's species mean distances to other species are kept.\n"
, "Species indicated with * will be removed from PFGs.\n"
, "Non-represented PFG might be one-species-only.")) +
.getGraphics_theme() +
theme(axis.ticks.x = element_blank()
, axis.text.x = element_text(angle = 90))
plot(pp3)
## GRAPHICAL REPRESENTATION 2 -----------------------------------------------
## Compute Principal Coordinates Analysis (PCO) for the determinantes of
## each group to see the position in "distance space" (overlap + traits)
## between species and groups
colRamp = colorRampPalette(c('#8e0152','#c51b7d','#de77ae','#7fbc41','#4d9221','#276419'))
pp4_list = foreach(group = group_names) %do%
{
## Transform results of clustering into distance matrix
tmp = determ[which(determ$GROUP == group),]
mat = mat.species.DIST[[group]]
mat = quasieuclid(as.dist(mat))
## Transform distance matrix into PCO
PCO = dudi.pco(mat, scannf = FALSE, nf = 3) ## PCO
PCO.li = PCO$li
PCO.li$det = ifelse(rownames(PCO.li) %in% tmp$sp[which(tmp$DETERMINANT == FALSE)], 0, 1)
PCO.li$det = factor(PCO.li$det, c(0, 1))
PCO.li$PFG = clust.groups[rownames(PCO.li)]
## GET inertia values
inert = inertia.dudi(PCO)$tot.inertia
inert = c(inert$`cum(%)`[1]
, inert$`cum(%)`[2] - inert$`cum(%)`[1]
, inert$`cum(%)`[3] - inert$`cum(%)`[2])
## GET ellipses
PCO.li.ELL = .getELLIPSE(xy = PCO.li[, c("A1", "A2")], fac = PCO.li$PFG)
PCO.li.ELL.det = .getELLIPSE(xy = PCO.li[which(PCO.li$det == 1), c("A1", "A2")]
, fac = PCO.li$PFG[which(PCO.li$det == 1)])
pp4 = ggplot(PCO.li, aes_string(x = "A1", y = "A2", color = "PFG")) +
geom_hline(yintercept = 0, color = "grey30", lwd = 1) +
geom_vline(xintercept = 0, color = "grey30", lwd = 1) +
geom_point(aes_string(shape = "det", size = "det"), alpha = 0.5) +
geom_path(data = PCO.li.ELL, aes_string(x = "x", y = "y"), lty = 2) +
geom_path(data = PCO.li.ELL.det, aes_string(x = "x", y = "y")) +
geom_label_repel(data = unique(PCO.li.ELL[, c("xlabel", "ylabel", "PFG")])
, aes_string(x = "xlabel", y = "ylabel", label = "PFG")) +
scale_shape_manual(guide = F, values = c("0" = 8, "1" = 20)) +
scale_size_manual(guide = F, values = c("0" = 3, "1" = 1)) +
scale_color_discrete(guide = F) +
labs(x = paste0("\n1st axis = ", round(inert[1], 1), "% of inertia")
, y = paste0("2nd axis = ", round(inert[2], 1), "% of inertia\n")
, title = paste0("STEP C : Removal of distant species : group ", group)
, subtitle = paste0("Only species whose mean distance to other species "
, "is included in the distribution\n"
, "of all PFG's species mean distances to other species are kept.\n"
, "Species indicated with * will be removed from PFGs.\n"
, "Inertia ellipse are represented, with (dashed) and "
, "without (solid) non-determinant species.")) +
.getGraphics_theme() +
theme(axis.title = element_text(inherit.blank = FALSE))
plot(pp4)
return(pp4)
}
names(pp4_list) = group_names
#############################################################################
cat("\n> Done!\n")
pdf(file = "PRE_FATE_CLUSTERING_STEP_2_distantSpecies_PCO.pdf"
, width = 10, height = 8)
plot(pp3)
for(group in group_names)
{
plot(pp4_list[[group]])
}
dev.off()
return(list(determ.sp = determ$species[which(determ$DETERMINANT == TRUE)]
, determ.all = determ
, plot.distance = pp3
, plot.PCO = pp4_list))
}
MayaGueguen/RFate documentation built on Oct. 17, 2020, 8:06 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions PRE_FATE.speciesClustering_step2
Documented in PRE_FATE.speciesClustering_step2
### HEADER #####################################################################
##' @title Choose clusters and select determinant species
##'
##' @name PRE_FATE.speciesClustering_step2
##'
##' @author Maya Guéguen
##'
##' @description This script is designed to obtain functional groups by : 1)
##' selecting the number of clusters to be kept from an object obtained with
##' the \code{\link{PRE_FATE.speciesClustering_step1}} function ; 2) refining
##' these groups by identifying determinant species in each of them.
##'
##' @param clust.dendrograms a dendrogram, or a \code{list} of dendrograms (one
##' for each \code{GROUP} value). \cr Such an object can be obtained
##' with the \code{\link{PRE_FATE.speciesClustering_step1}} function.
##' @param no.clusters an \code{integer}, or a \code{vector} of \code{integer}
##' (one for each \code{GROUP} value), with the number of clusters to be kept
##' @param mat.species.DIST a \code{dist} object, or a \code{list} of
##' \code{dist} objects (one for each \code{GROUP} value), corresponding to the
##' distance between each pair of species. \cr Such an object can be obtained
##' with the \code{\link{PRE_FATE.speciesDistance}} function.
##'
##'
##' @details
##'
##' This function allows to obtain a classification of \strong{dominant}
##' species into Plant Functional Groups (PFG), and the \strong{determinant}
##' species based on these PFG. \cr \cr
##'
##' \strong{What is the difference between \code{dominant} and
##' \code{determinant} species ?}
##'
##' \itemize{
##' \item{\strong{Dominant} species are species representative of an
##' environment or a studied area, in terms of number of releves or
##' abundance values. They can be found with the
##' \code{\link{PRE_FATE.selectDominant}} function of this package. These
##' dominant species are used to build PFG with the
##' \code{\link{PRE_FATE.speciesClustering_step1}} function.
##' }
##' \item{Once PFG are built, \strong{determinant} species are defined as
##' refined subsets of dominant species within each PFG. \cr The process is
##' detailed below :
##' \itemize{
##' \item each dominant species is assigned to a PFG
##' \item within each PFG :
##' \itemize{
##' \item for each species, compute its mean distance to the other
##' species within the PFG (\code{sp.mean.dist})
##' \item calculate the mean value of all these mean distances
##' (\code{allSp.mean})
##' \item calculate the deviation values around this mean value
##' (\code{allSp.min} and \code{allSp.max})
##' \item determinant species are the ones that are included between
##' those deviation values
##' }
##' }
##' }
##' }
##'
##' @return A \code{list} containing one \code{vector}, one \code{data.frame}
##' with the following columns, and two \code{ggplot2} objects :
##'
##' \describe{
##' \item{determ.sp}{the names of all determinant species}
##' \item{determ.all}{(\emph{determinant and non-determinant species})
##' \describe{
##' \item{\code{PFG}}{ID of the plant functional group
##' (\code{GROUP} + \code{ID.cluster})}
##' \item{\code{GROUP}}{name of data subset}
##' \item{\code{ID.cluster}}{cluster number}
##' \item{\code{species}}{name of species}
##' \item{\code{ID.species}}{species number in each PFG}
##' \item{\code{sp.mean.dist}}{species mean distance to other species of
##' the same PFG}
##' \item{\code{allSp.mean}}{\eqn{mean(\text{sp.mean.dist})} within the PFG}
##' \item{\code{allSp.min}}{\eqn{mean(\text{sp.mean.dist}) - 1.64 *
##' sd(\text{sp.mean.dist})} within the PFG}
##' \item{\code{allSp.max}}{\eqn{mean(\text{sp.mean.dist}) + 1.64 *
##' sd(\text{sp.mean.dist})} within the PFG}
##' \item{\code{DETERMINANT}}{\code{TRUE} if determinant species, \code{FALSE}
##' otherwise}
##' }
##' }
##' \item{plot.distance}{\code{ggplot2} object, representing the distribution
##' of mean distances between species for each functional group}
##' \item{plot.PCO}{\code{list} of \code{ggplot2} objects, representing the
##' PFG within the functional space \cr \cr}
##' }
##'
##' One \file{PRE_FATE_CLUSTERING_STEP_2_distantSpecies_PCO.pdf} file is created
##' containing two types of graphics :
##' \describe{
##' \item{distantSpecies}{to visualize in each PFG the distribution of mean
##' distance of each species to other species, and non-determinant species
##' which are outside the distribution}
##' \item{PCO}{to visualize in each PFG the distribution of species, with and
##' without non-determinant species}
##' }
##'
##' @keywords hierarchical clustering, Principal Component Ordination
##'
##' @seealso \code{\link[stats]{cutree}},
##' \code{\link[ade4]{quasieuclid}},
##' \code{\link[ade4]{dudi.pco}},
##' \code{\link{PRE_FATE.speciesDistance}},
##' \code{\link{PRE_FATE.speciesClustering_step1}},
##' \code{\link{PRE_FATE.speciesClustering_step3}}
##'
##' @examples
##'
##' ## Load example data
##' data(DATASET_Bauges_PFG)
##'
##' ## Species dissimilarity distance (niche overlap + traits distance)
##' tab.dist = DATASET_Bauges_PFG$dom.dist_total
##' str(tab.dist)
##' as.matrix(tab.dist[[1]])[1:5, 1:5]
##'
##'
##' ## Build dendrograms -------------------------------------------------------------------------
##' sp.CLUST = PRE_FATE.speciesClustering_step1(mat.species.DIST = tab.dist)
##' names(sp.CLUST)
##'
##'
##' ## Number of clusters per group
##' plot(sp.CLUST$plot.clustNo)
##' no.clusters = c(5, 5, 9)
##'
##' ## Find determinant species ------------------------------------------------------------------
##' sp.DETERM = PRE_FATE.speciesClustering_step2(clust.dendrograms = sp.CLUST$clust.dendrograms
##' , no.clusters = no.clusters
##' , mat.species.DIST = tab.dist)
##' names(sp.DETERM)
##'
##' str(sp.DETERM$determ.sp)
##' str(sp.DETERM$determ.all)
##'
##' ## Species names
##' sp.NAMES = DATASET_Bauges_PFG$sp.names
##' sp.NAMES$species = paste0("X", sp.NAMES$species)
##' determ = merge(sp.DETERM$determ.all, sp.NAMES, by = "species", all.x = TRUE)
##' str(determ)
##'
##' plot(sp.DETERM$plot.distance)
##' plot(sp.DETERM$plot.PCO$Chamaephyte)
##' plot(sp.DETERM$plot.PCO$Herbaceous)
##' plot(sp.DETERM$plot.PCO$Phanerophyte)
##'
##'
##' @export
##'
##' @importFrom grDevices colorRampPalette pdf dev.off
##' @importFrom stats as.dist cutree sd
##' @importFrom graphics plot
##'
##' @importFrom ade4 quasieuclid dudi.pco inertia.dudi
##'
##' @importFrom ggplot2 ggplot ggsave aes_string
##' geom_point geom_hline geom_vline geom_errorbar geom_path
##' scale_color_discrete scale_color_manual scale_shape_manual scale_size_manual
##' facet_grid labs theme element_text element_blank
##' @importFrom ggthemes theme_fivethirtyeight
##' @importFrom ggrepel geom_label_repel
##'
## END OF HEADER ###############################################################
PRE_FATE.speciesClustering_step2 = function(clust.dendrograms
, no.clusters
, mat.species.DIST
){
#############################################################################
## CHECK parameter clust.dendrograms
if (.testParam_notInClass(clust.dendrograms, c("list","hclust")))
{
stop("No data given!\n (missing `clust.dendrograms` information which must be of class `hclust` or a list `hclust` objects)")
} else
{
if (class(clust.dendrograms) == "list" &&
length(which(sapply(clust.dendrograms, class) == "hclust")) < length(clust.dendrograms))
{
stop("Wrong type of data!\n each element of `clust.dendrograms` must be of class `hclust`")
}
if (class(clust.dendrograms) == "hclust")
{
clust.dendrograms = list(GROUP1 = clust.dendrograms)
}
if(!is.null(names(clust.dendrograms)))
{
group_names = names(clust.dendrograms)
} else
{
group_names = paste0("GROUP", 1:length(clust.dendrograms))
}
}
## CHECK parameter no.clusters
if (.testParam_notNum(no.clusters))
{
stop("No data given!\n (missing `no.clusters` information)")
} else
{
if (length(no.clusters) != length(clust.dendrograms))
{
stop("Wrong type of data!\n `no.clusters` must have the same length than `clust.dendrograms`")
}
}
## CHECK parameter mat.species.DIST
if (.testParam_notInClass(mat.species.DIST, c("list","dist")))
{
stop("No data given!\n (missing `mat.species.DIST` information which must be a dist object, or a list of dist objects)")
} else
{
if (class(mat.species.DIST) == "list" &&
length(mat.species.DIST) != length(clust.dendrograms))
{
stop("Wrong type of data!\n `mat.species.DIST` must have the same length than `clust.dendrograms`")
}
if (class(mat.species.DIST) == "dist")
{
mat.species.DIST = list(GROUP1 = mat.species.DIST)
}
}
cat("\n\n #------------------------------------------------------------#")
cat("\n # PRE_FATE.speciesClustering_step2 : DETERMINANT SPECIES")
cat("\n #------------------------------------------------------------# \n")
#############################################################################
## DEFINITION OF CLUSTERED GROUPS
#############################################################################
### CUT dendrogramS (or trees) RESULTING FROM hclust INTO SEVERAL GROUPS (no = k)
## Number of groups for each group has been chosen
## according to the previous plot (CLUSTERING_STEP_1B)
clust.groups = lapply(1:length(group_names), function(x)
cutree(clust.dendrograms[[x]], k = no.clusters[x]))
clust.groups = lapply(1:length(group_names), function(x) {
tmp.names = names(clust.groups[[x]])
tmp = paste0(group_names[x], ".", clust.groups[[x]])
names(tmp) = tmp.names
return(tmp)
})
clust.groups = unlist(clust.groups)
PFG_names = sort(unique(clust.groups))
#############################################################################
## IDENTIFY DETERMINANT SPECIES
#############################################################################
## DETERMINANT SPECIES are the ones whose mean distance to other species
## is in the distribution of mean distances of every species to other species
determ = foreach(pfg = PFG_names) %do%
{
## get the SPECIES names
sp = names(clust.groups)[which(clust.groups == pfg)]
if (length(sp) > 1)
{
## get the GROUP information
group = strsplit(pfg, "[.]")[[1]][1]
ID.cluster = strsplit(pfg, "[.]")[[1]][2]
## get the DISTANCE information
mat = as.matrix(mat.species.DIST[[group]])[sp, sp]
## compute for each species its mean distance to other species
sp.mean.dist = apply(mat, 1, mean, na.rm = T)
return(data.frame(PFG = pfg
, GROUP = group
, ID.cluster
, species = sp
, ID = 1:length(sp)
, sp.mean.dist
, allSp.mean = mean(sp.mean.dist)
, allSp.min = mean(sp.mean.dist) - 1.64 * sd(sp.mean.dist)
, allSp.max = mean(sp.mean.dist) + 1.64 * sd(sp.mean.dist)
, stringsAsFactors = FALSE))
} else if (length(sp) == 1)
{
return(data.frame(PFG = pfg
, GROUP = group
, ID.cluster
, species = sp
, ID = 1
, sp.mean.dist = NA
, allSp.mean = NA
, allSp.min = NA
, allSp.max = NA
, stringsAsFactors = FALSE))
}
}
determ = do.call(rbind, determ)
rownames(determ) = determ$species
if (length(determ) == 0 || is.null(determ))
{
stop("No determinant species have been selected. Please check your data")
}
determ$DETERMINANT = TRUE
determ$DETERMINANT[which(determ$sp.mean.dist > determ$allSp.max)] = FALSE
determ$DETERMINANT[which(determ$sp.mean.dist < determ$allSp.min)] = FALSE
determ$DETERMINANT = factor(determ$DETERMINANT, c(TRUE, FALSE))
## SAVE DETERM
## CAT INFO ABOUT HOW MANY SPECIES ARE REMOVED
## CAT INFO ABOUT HOW MANY SPECIES in each PFG
#############################################################################
## GRAPHICAL REPRESENTATIONS
#############################################################################
## GRAPHICAL REPRESENTATION 1 -----------------------------------------------
colRamp = colorRampPalette(c('#8e0152','#c51b7d','#de77ae','#7fbc41','#4d9221','#276419'))
colLev = levels(interaction(determ$DETERMINANT, determ$GROUP))
pp3 = ggplot(determ, aes_string(x = "PFG", y = "sp.mean.dist"
, color = interaction(determ$DETERMINANT, determ$GROUP)
, shape = "DETERMINANT")) +
scale_color_manual(guide = F, values = colRamp(length(colLev))) +
scale_shape_manual(guide = F, values = c("0" = 20, "1" = 8)) +
geom_errorbar(aes_string(ymin = "allSp.min", ymax = "allSp.max")
, color = "darkblue") +
geom_point(position = "jitter") +
geom_point(aes_string(y = "allSp.mean")
, pch = 18
, lwd = 5
, color = "darkblue") +
facet_grid("~ GROUP", scales = "free_x") +
labs(x = "", y = "Mean distance to other species"
, title = "STEP C : Removal of distant species"
, subtitle = paste0("Only species whose mean distance to other species "
, "is included in the distribution\n"
, "of all PFG's species mean distances to other species are kept.\n"
, "Species indicated with * will be removed from PFGs.\n"
, "Non-represented PFG might be one-species-only.")) +
.getGraphics_theme() +
theme(axis.ticks.x = element_blank()
, axis.text.x = element_text(angle = 90))
plot(pp3)
## GRAPHICAL REPRESENTATION 2 -----------------------------------------------
## Compute Principal Coordinates Analysis (PCO) for the determinantes of
## each group to see the position in "distance space" (overlap + traits)
## between species and groups
colRamp = colorRampPalette(c('#8e0152','#c51b7d','#de77ae','#7fbc41','#4d9221','#276419'))
pp4_list = foreach(group = group_names) %do%
{
## Transform results of clustering into distance matrix
tmp = determ[which(determ$GROUP == group),]
mat = mat.species.DIST[[group]]
mat = quasieuclid(as.dist(mat))
## Transform distance matrix into PCO
PCO = dudi.pco(mat, scannf = FALSE, nf = 3) ## PCO
PCO.li = PCO$li
PCO.li$det = ifelse(rownames(PCO.li) %in% tmp$sp[which(tmp$DETERMINANT == FALSE)], 0, 1)
PCO.li$det = factor(PCO.li$det, c(0, 1))
PCO.li$PFG = clust.groups[rownames(PCO.li)]
## GET inertia values
inert = inertia.dudi(PCO)$tot.inertia
inert = c(inert$`cum(%)`[1]
, inert$`cum(%)`[2] - inert$`cum(%)`[1]
, inert$`cum(%)`[3] - inert$`cum(%)`[2])
## GET ellipses
PCO.li.ELL = .getELLIPSE(xy = PCO.li[, c("A1", "A2")], fac = PCO.li$PFG)
PCO.li.ELL.det = .getELLIPSE(xy = PCO.li[which(PCO.li$det == 1), c("A1", "A2")]
, fac = PCO.li$PFG[which(PCO.li$det == 1)])
pp4 = ggplot(PCO.li, aes_string(x = "A1", y = "A2", color = "PFG")) +
geom_hline(yintercept = 0, color = "grey30", lwd = 1) +
geom_vline(xintercept = 0, color = "grey30", lwd = 1) +
geom_point(aes_string(shape = "det", size = "det"), alpha = 0.5) +
geom_path(data = PCO.li.ELL, aes_string(x = "x", y = "y"), lty = 2) +
geom_path(data = PCO.li.ELL.det, aes_string(x = "x", y = "y")) +
geom_label_repel(data = unique(PCO.li.ELL[, c("xlabel", "ylabel", "PFG")])
, aes_string(x = "xlabel", y = "ylabel", label = "PFG")) +
scale_shape_manual(guide = F, values = c("0" = 8, "1" = 20)) +
scale_size_manual(guide = F, values = c("0" = 3, "1" = 1)) +
scale_color_discrete(guide = F) +
labs(x = paste0("\n1st axis = ", round(inert[1], 1), "% of inertia")
, y = paste0("2nd axis = ", round(inert[2], 1), "% of inertia\n")
, title = paste0("STEP C : Removal of distant species : group ", group)
, subtitle = paste0("Only species whose mean distance to other species "
, "is included in the distribution\n"
, "of all PFG's species mean distances to other species are kept.\n"
, "Species indicated with * will be removed from PFGs.\n"
, "Inertia ellipse are represented, with (dashed) and "
, "without (solid) non-determinant species.")) +
.getGraphics_theme() +
theme(axis.title = element_text(inherit.blank = FALSE))
plot(pp4)
return(pp4)
}
names(pp4_list) = group_names
#############################################################################
cat("\n> Done!\n")
pdf(file = "PRE_FATE_CLUSTERING_STEP_2_distantSpecies_PCO.pdf"
, width = 10, height = 8)
plot(pp3)
for(group in group_names)
{
plot(pp4_list[[group]])
}
dev.off()
return(list(determ.sp = determ$species[which(determ$DETERMINANT == TRUE)]
, determ.all = determ
, plot.distance = pp3
, plot.PCO = pp4_list))
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.