add edge_node_distance class

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: meconetcomp
 Type: Package
 Title: Compare Microbial Networks of 'trans_network' Class of 'microeco' Package
-Version: 0.1.1
+Version: 0.2.0
 Authors@R: c(person("Chi", "Liu", role = c("aut", "cre"), email = "liuchi0426@126.com"),
               person("Minjie", "Yao", role = "ctb", email = "yaomj@fafu.edu.cn"),
               person("Xiangzhen", "Li", role = "ctb", email = "lixz@cib.ac.cn")
@@ -12,13 +12,13 @@ Description: Compare microbial co-occurrence networks created from 'trans_networ
     This package is the extension of 'trans_network' class of 'microeco' package and especially useful when different networks are constructed and analyzed simultaneously.
 URL: https://github.com/ChiLiubio/meconetcomp
 Depends: R (>= 3.5.0)
-Imports: microeco, magrittr, dplyr, igraph, reshape2
-Suggests: rgexf, ape, file2meco
+Imports: R6, microeco (>= 0.12.0), magrittr, dplyr, igraph, reshape2, ggpubr
+Suggests: rgexf, ape, file2meco, agricolae
 License: GPL-3
 LazyData: true
 Encoding: UTF-8
 NeedsCompilation: no
-Packaged: 2022-09-04
+Packaged: 2022-10-04
 Repository: CRAN
-Date/Publication: 2022-09-07
-RoxygenNote: 7.1.2
+Date/Publication: 2022-10-04
+RoxygenNote: 7.2.1

NAMESPACE

@@ -3,13 +3,15 @@
 export(cal_module)
 export(cal_network_attr)
 export(edge_comp)
+export(edge_node_distance)
 export(edge_tax_comp)
 export(get_edge_table)
 export(get_node_table)
 export(node_comp)
 export(subnet_property)
 export(subset_network)
 import(microeco)
+importFrom(R6,R6Class)
 importFrom(dplyr,full_join)
 importFrom(igraph,is_directed)
 importFrom(magrittr,"%<>%")

R/cal_module.R

@@ -3,12 +3,12 @@
 #' @description
 #' Calculating modularity of networks and assign the modules to nodes for each network.
 #'
-#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of microeco package.
+#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of \code{microeco} package.
 #' @param undirected_method default "cluster_fast_greedy"; the modularity algorithm for undirected network; 
-#'       see cal_module function of \code{\link{trans_network}} class for more algorithms.
+#'       see \code{cal_module} function of \code{\link{trans_network}} class for more algorithms.
 #' @param directed_method default 'cluster_optimal'; the modularity algorithm for directed network.
-#' @param ... other parameters (except for method) passed to cal_module function of \code{\link{trans_network}} class.
-#' @return list, with module attribute in nodes of each network
+#' @param ... other parameters (except for method) passed to \code{cal_module} function of \code{\link{trans_network}} class.
+#' @return \code{list}, with module attribute in nodes of each network
 #' @examples
 #' data(soil_amp_network)
 #' soil_amp_network <- cal_module(soil_amp_network)

R/cal_network_attr.R

@@ -3,8 +3,8 @@
 #' @description
 #' Calculate the topological properties of all the networks and merge the results into one table.
 #'
-#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of microeco package.
-#' @return data.frame
+#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of \code{microeco} package.
+#' @return \code{data.frame}
 #' @examples
 #' data(soil_amp_network)
 #' test <- cal_network_attr(soil_amp_network)

R/edge_comp.R

@@ -1,11 +1,11 @@
-#' Generate a microtable object with paired nodes distributions of edges across networks
+#' Generate a \code{microtable} object with paired nodes distributions of edges across networks
 #'
 #' @description
-#' Generate a microtable object with paired nodes distributions of edges across networks. Useful for the edge comparisons across different networks.
-#' The return otu_table in microtable object has the binary numbers in which 1 represents the presence of the edge in the corresponding network.
+#' Generate a \code{microtable} object with paired nodes distributions of edges across networks. Useful for the edge comparisons across different networks.
+#' The return \code{otu_table} in \code{microtable} object has the binary numbers in which 1 represents the presence of the edge in the corresponding network.
 #'
 #' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of microeco package.
-#' @return microtable object
+#' @return \code{microtable} object
 #' @examples
 #' data(soil_amp_network)
 #' test <- edge_comp(soil_amp_network)

R/edge_node_distance.R

@@ -0,0 +1,174 @@
+#' @title Perform the distance distribution of paired nodes in edges across networks.
+#'
+#' @description
+#' This class is a wrapper for a series of analysis on the distance distribution 
+#' of paired nodes in edges across networks and the differential test.
+#'
+#' @export
+edge_node_distance <- R6::R6Class(classname = "edge_node_distance",
+	public = list(
+		#' @param network_list a list with multiple networks; all the networks should be \code{trans_network} object 
+		#' 	 created from \code{\link{trans_network}} class of microeco package.
+		#' @param dis_matrix default NULL; the distance matrix of nodes, used for the value extraction; 
+		#' 	 must be a symmetrical matrix with both colnames and rownames.
+		#' @param label default "+"; "+" or "-" or \code{c("+", "-")}; the edge label used for the selection of edges.
+		#' @param with_module default FALSE; whether show the module classification of nodes in the result.
+		#' @param module_thres default 2; the threshold of the nodes number of modules remained when \code{with_module = TRUE}.
+		#' @return \code{data_table} in the object
+		#' @examples
+		#' \donttest{
+		#' data(soil_amp_network)
+		#' data(soil_amp)
+		#' # select a small dataset to speed up the calculation
+		#' node_names <- unique(unlist(lapply(soil_amp_network, function(x){colnames(x$data_abund)})))
+		#' filter_soil_amp <- microeco::clone(soil_amp)
+		#' filter_soil_amp$otu_table <- filter_soil_amp$otu_table[node_names, ]
+		#' filter_soil_amp$tidy_dataset()
+		#' # obtain phylogenetic distance matrix
+		#' phylogenetic_distance <- as.matrix(cophenetic(filter_soil_amp$phylo_tree))
+		#' # choose the positive labels
+		#' t1 <- edge_node_distance$new(network_list = soil_amp_network, dis_matrix = phylogenetic_distance, label = "+")
+		#' }
+		initialize = function(network_list, dis_matrix = NULL, label = "+", with_module = FALSE, module_thres = 2){
+			check_input(network_list)
+			if(is.null(dis_matrix)){
+				stop("Please provide dis_matrix parameter!")
+			}
+			res_table <- data.frame()
+			for(i in names(network_list)){
+				tmp <- private$get_matrix_value(
+					network = network_list[[i]], 
+					label = label, 
+					dis_matrix = dis_matrix, 
+					group_name = i, 
+					with_module = with_module, 
+					module_thres = module_thres
+					)
+				res_table %<>% rbind(., tmp)
+			}
+			res_table$Group %<>% factor(., levels = names(network_list))
+			res_table %<>% .[!is.na(.$Value), ]
+			self$data_table <- res_table
+			self$label <- label
+		},
+		#' @description
+		#' Differential test across networks.
+		#'
+		#' @param method default "KW"; see the following available options:
+		#'   \describe{
+		#'     \item{\strong{'anova'}}{Duncan's multiple range test for anova}
+		#'     \item{\strong{'KW'}}{KW: Kruskal-Wallis Rank Sum Test for all groups (>= 2)}
+		#'     \item{\strong{'KW_dunn'}}{Dunn's Kruskal-Wallis Multiple Comparisons, see \code{dunnTest} function in \code{FSA} package}
+		#'     \item{\strong{'wilcox'}}{Wilcoxon Rank Sum and Signed Rank Tests for all paired groups }
+		#'     \item{\strong{'t.test'}}{Student's t-Test for all paired groups}
+		#'   }
+		#' @param ... parameters passed to \code{cal_diff} function of \code{trans_alpha} class of \code{microeco} package.
+		#' @return \code{res_diff} in object. See the Return of \code{cal_diff} function in \code{trans_alpha} class of \code{microeco} package.
+		#' @examples
+		#' \donttest{
+		#' t1$cal_diff(method = "wilcox")
+		#' }
+		cal_diff = function(method = c("anova", "KW", "KW_dunn", "wilcox", "t.test")[1], ...){
+			res <- self$data_table
+			res$Measure <- "Value"
+			# two cases: only one type of label and two types of labels
+			if(length(unique(res$label)) == 1){
+				suppressMessages(tmp2 <- trans_alpha$new(dataset = NULL))
+				tmp2$data_alpha <- res
+				tmp2$group <- "Group"
+				tmp2$cal_diff(method = method, measure = "Value", ...)
+			}else{
+				res$raw_Group <- res$Group
+				res$Label <- paste0(res$Group, " - ", res$label)
+				suppressMessages(tmp2 <- trans_alpha$new(dataset = NULL))
+				tmp2$data_alpha <- res
+				tmp2$group <- "Label"
+				if(method != "anova"){
+					message("For multiple labels, only anova can be used!")
+				}
+				tmp2$cal_diff(method = "anova", measure = "Value", ...)
+				split_raw <- strsplit(rownames(tmp2$res_diff), split = " - ")
+				tmp2$res_diff$by_group <- lapply(split_raw, function(x){x[1]}) %>% unlist
+				tmp2$res_diff$Group <- lapply(split_raw, function(x){x[2]}) %>% unlist
+				res$by_group <- res$raw_Group
+				res$Label <- res$label
+				res$Label %<>% factor(., levels = self$label)
+				tmp2$data_alpha <- res
+				tmp2$by_group <- "by_group"
+			}
+			self$res_diff <- tmp2$res_diff
+			self$cal_diff_method <- method
+			self$tmp_diff <- tmp2
+			message('The result is stored in object$res_diff ...')
+		},
+		#' @description
+		#' Plot the distance.
+		#'
+		#' @param ... parameters pass to \code{plot_alpha} function of \code{trans_alpha} class of \code{microeco} package.
+		#' @return \code{ggplot}.
+		#' @examples
+		#' \donttest{
+		#' t1$plot(boxplot_add = "none", add_sig = TRUE)
+		#' }
+		plot = function(
+			...
+			){
+			self$tmp_diff$plot_alpha(measure = "Value", ...)
+		}
+		),
+	private = list(
+		get_matrix_value = function(network, label, dis_matrix, group_name, with_module, module_thres){
+			if(!with_module){
+				if(is.null(network$res_edge_table)){
+					network$get_edge_table()
+				}
+				tmp <- network$res_edge_table %>% .[.$label %in% label, ]
+				if(nrow(tmp) == 0){
+					res <- NA
+				}else{
+					select_value <- private$select_value_matrix(input_table = tmp, input_matrix = dis_matrix)
+					res <- data.frame(Group = group_name, Value = select_value, label = tmp$label)
+				}
+			}else{
+				if(!is.numeric(module_thres)){
+					stop("module_thres must be numeric!")
+				}
+				if(! "module" %in% colnames(network$res_node_table)){
+					stop("please first use cal_module function to calculate modularity!")
+				}
+				if(is.null(network$res_node_table)){
+					network$get_node_table(node_roles = FALSE)
+				}
+				# check module nodes number
+				use_modules <- table(network$res_node_table$module) %>% .[. >= module_thres] %>% names
+				res <- NULL
+				for(k in use_modules){
+					module_nodes <- network$res_node_table %>% .[.$module == k, ] %>% rownames
+					t1 <- clone(network)
+					t1$res_network <- t1$subset_network(node = module_nodes, rm_single = TRUE)
+					suppressMessages(t1$get_edge_table())
+					tmp <- t1$res_edge_table %>% .[.$label %in% label, ]
+					if(nrow(tmp) == 0){
+						next
+					}else{
+						select_value <- private$select_value_matrix(input_table = tmp, input_matrix = dis_matrix)
+						res <- rbind(res, data.frame(Group = group_name, Value = select_value, module = k, label = tmp$label))
+					}
+				}
+			}
+			res
+		},
+		select_value_matrix = function(input_table, input_matrix){
+			select_value <- lapply(seq_len(nrow(input_table)), function(x){
+				if(all(c(input_table[x, 1], input_table[x, 2]) %in% colnames(input_matrix))){
+					input_matrix[input_table[x, 1], input_table[x, 2]]
+				}else{
+					NA
+				}
+			}) %>% unlist
+			select_value
+		}
+	),
+	lock_objects = FALSE,
+	lock_class = FALSE
+)

R/edge_tax_comp.R

@@ -3,11 +3,12 @@
 #' @description
 #' Taxonomic sum of linked nodes in edges across networks.
 #'
-#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of microeco package.
+#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of \code{microeco} package.
 #' @param taxrank default "Phylum"; Which taxonomic level is used for the sum of nodes in edges.
-#' @param label default "+"; "+" or "-" or c("+", "-"); the edge label used for the selection of edges for the sum.
-#' @param rel default TRUE; TRUE represents using ratio, the denominator is the number of selected edges; FALSE represents the absolute number of the sum of edges.
-#' @return data.frame
+#' @param label default "+"; "+" or "-" or \code{c("+", "-")}; the edge label used for the selection of edges for the sum.
+#' @param rel default \code{TRUE}; \code{TRUE} represents using ratio, the denominator is the number of selected edges; 
+#'   \code{FALSE} represents the absolute number of the sum of edges.
+#' @return \code{data.frame}
 #' @examples
 #' data(soil_amp_network)
 #' test <- edge_tax_comp(soil_amp_network)

R/get_edge_table.R

@@ -3,8 +3,8 @@
 #' @description
 #' Get edge property table for each network in the list with multiple networks.
 #'
-#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of microeco package.
-#' @return list, with res_edge_table in each network
+#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of \code{microeco} package.
+#' @return \code{list}, with \code{res_edge_table} in each network
 #' @examples
 #' data(soil_amp_network)
 #' soil_amp_network <- get_edge_table(soil_amp_network)

R/get_node_table.R

@@ -3,9 +3,9 @@
 #' @description
 #' Get node property table for each network in the list with multiple networks.
 #'
-#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of microeco package.
+#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of \code{microeco} package.
 #' @param ... parameter passed to get_node_table function of \code{\link{trans_network}} class.
-#' @return list, with res_node_table in each network
+#' @return \code{list}, with \code{res_node_table} in each network
 #' @examples
 #' \donttest{
 #' data(soil_amp_network)

R/import.R

@@ -1,4 +1,5 @@
 #' @import microeco
+#' @importFrom R6 R6Class
 #' @importFrom magrittr %<>%
 #' @importFrom magrittr %>%
 #' @importFrom dplyr full_join

R/node_comp.R

@@ -3,12 +3,12 @@
 #' @description
 #' Generate a microtable object with node distributions across networks. Useful for the node information comparisons across different networks.
 #' 
-#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of microeco package.
-#' @param property default "name"; a colname of res_node_table in each network; 
+#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of \code{microeco} package.
+#' @param property default "name"; a colname of \code{res_node_table} in each network; 
 #'    the default "name" represents using node presence/absence information in the otu_table of final output, in which
 #'    1 represents presence of the node in the corresponding network; 
-#'    For other options (such as degree), the results in the output otu_table are the actual values of res_node_table.
-#' @return microtable object
+#'    For other options (such as degree), the results in the output otu_table are the actual values of \code{res_node_table}.
+#' @return \code{microtable} object
 #' @examples
 #' \donttest{
 #' data(soil_amp_network)

R/subnet_property.R

@@ -3,8 +3,8 @@
 #' @description
 #' Extracting sub-network according to the presence of features in each sample across networks and calculate the sub-network properties.
 #' 
-#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of microeco package.
-#' @return data.frame
+#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of \code{microeco} package.
+#' @return \code{data.frame}
 #' @examples
 #' \donttest{
 #' data(soil_amp_network)

R/subset_network.R

@@ -3,10 +3,10 @@
 #' @description
 #' Extracting a network according to the edge intersection of networks.
 #' 
-#' @param network_list a list with multiple networks; all the networks should be trans_network object created from \code{\link{trans_network}} class of microeco package.
-#' @param venn default NULL; a microtable object which must be converted by trans_comm function of trans_venn class.
-#' @param name default NULL; integer or character; must be a number or one of colnames of the otu_table in venn.
-#' @return a trans_network object, with only the extracted edges in the network
+#' @param network_list a list with multiple networks; all the networks should be \code{trans_network} object created from \code{\link{trans_network}} class of microeco package.
+#' @param venn default NULL; a \code{microtable} object which must be converted by \code{trans_comm} function of \code{trans_venn} class.
+#' @param name default NULL; integer or character; must be a number or one of colnames of the \code{otu_table} in the input \code{venn} parameter.
+#' @return \code{trans_network} object, with only the extracted edges in the network
 #' @examples
 #' \donttest{
 #' data(soil_amp_network)