Add SEA prediction module

.Rbuildignore

@@ -3,3 +3,4 @@
 Makefile
 README.Rmd
 README_files
+^data-raw$

.gitignore

@@ -1 +1,6 @@
 .Rproj.user
+.DS_Store
+.RData
+.Rhistory
+*.RData
+*.Rhistory

DESCRIPTION

@@ -7,11 +7,36 @@ Description: What the package does (one paragraph).
 Depends:
     R (>= 3.5.0)
 Imports:
+    aplot,
+    circlize,
     dplyr,
+    enrichplot,
+    forcats,
+    ggalluvial,
+    ggfun,
+    ggplot2,
+    ggrepel,
+    ggvenn,
+    grDevices,
     igraph,
+    magrittr,
+    MCL,
+    methods,
+    paletteer,
+    purrr,
+    RColorBrewer,
     rlang,
-    tidyr
+    scales,
+    stats,
+    tibble,
+    tidyr,
+    utils,
+    yulab.utils
+Suggests:
+    ComplexHeatmap,
+    testthat
 License: Artistic-2.0
 Encoding: UTF-8
 Roxygen: list(markdown = TRUE)
-RoxygenNote: 7.3.2
+RoxygenNote: 7.3.3
+LazyData: true

NAMESPACE

@@ -1,26 +1,141 @@
 # Generated by roxygen2: do not edit by hand
 
+export(Addclusterscore)
+export(PlotNodeHeatmap)
+export(TCM_sankey)
+export(compute_nodeinfo)
+export(getGores)
 export(getMCODE_res)
+export(get_node_profile)
+export(ggdock)
+export(ggdot_sankey)
+export(gglollipop)
+export(ggvenn_plot)
+export(gocircle_plot)
+export(ppi_subset)
+export(radar_plot)
+export(rank_ppi_nodes)
 export(runMCODE)
+export(run_MCL)
+export(run_louvain)
 export(search_herb)
 export(search_target)
+import(ggplot2)
+importFrom(MCL,mcl)
+importFrom(RColorBrewer,brewer.pal)
+importFrom(RColorBrewer,brewer.pal.info)
+importFrom(aplot,plot_list)
+importFrom(circlize,circos.axis)
+importFrom(circlize,circos.clear)
+importFrom(circlize,circos.genomicInitialize)
+importFrom(circlize,circos.genomicRect)
+importFrom(circlize,circos.genomicTrack)
+importFrom(circlize,circos.genomicTrackPlotRegion)
+importFrom(circlize,circos.par)
+importFrom(circlize,circos.text)
+importFrom(circlize,circos.track)
+importFrom(circlize,colorRamp2)
+importFrom(circlize,get.cell.meta.data)
 importFrom(dplyr,"%>%")
+importFrom(dplyr,all_of)
+importFrom(dplyr,arrange)
+importFrom(dplyr,bind_rows)
+importFrom(dplyr,case_when)
+importFrom(dplyr,count)
+importFrom(dplyr,desc)
 importFrom(dplyr,distinct)
 importFrom(dplyr,filter)
+importFrom(dplyr,group_by)
+importFrom(dplyr,lead)
+importFrom(dplyr,left_join)
+importFrom(dplyr,mutate)
+importFrom(dplyr,pull)
 importFrom(dplyr,select)
+importFrom(dplyr,setdiff)
+importFrom(dplyr,slice)
+importFrom(dplyr,slice_head)
+importFrom(dplyr,summarise)
+importFrom(dplyr,ungroup)
+importFrom(enrichplot,dotplot)
+importFrom(forcats,fct_reorder)
+importFrom(ggalluvial,StatStratum)
+importFrom(ggalluvial,geom_flow)
+importFrom(ggalluvial,geom_stratum)
+importFrom(ggalluvial,to_lodes_form)
+importFrom(ggfun,get_legend)
+importFrom(ggplot2,coord_equal)
+importFrom(ggplot2,element_text)
+importFrom(ggplot2,expansion)
+importFrom(ggplot2,geom_polygon)
+importFrom(ggplot2,geom_segment)
+importFrom(ggplot2,geom_text)
+importFrom(ggplot2,ggplot)
+importFrom(ggplot2,ggtitle)
+importFrom(ggplot2,margin)
+importFrom(ggplot2,scale_x_continuous)
+importFrom(ggplot2,theme)
+importFrom(ggplot2,theme_void)
+importFrom(ggrepel,geom_text_repel)
+importFrom(ggvenn,ggvenn)
+importFrom(grDevices,colorRampPalette)
+importFrom(graphics,par)
+importFrom(grid,gpar)
+importFrom(grid,unit)
 importFrom(igraph,"V<-")
+importFrom(igraph,E)
 importFrom(igraph,V)
+importFrom(igraph,articulation_points)
+importFrom(igraph,as_adj_list)
+importFrom(igraph,as_adjacency_matrix)
 importFrom(igraph,as_data_frame)
 importFrom(igraph,as_undirected)
+importFrom(igraph,betweenness)
+importFrom(igraph,bfs)
+importFrom(igraph,closeness)
+importFrom(igraph,cluster_louvain)
+importFrom(igraph,components)
 importFrom(igraph,coreness)
 importFrom(igraph,degree)
 importFrom(igraph,delete_vertices)
+importFrom(igraph,distances)
+importFrom(igraph,eccentricity)
+importFrom(igraph,ecount)
+importFrom(igraph,edge_attr)
+importFrom(igraph,edge_attr_names)
 importFrom(igraph,edge_density)
+importFrom(igraph,eigen_centrality)
 importFrom(igraph,induced_subgraph)
 importFrom(igraph,is_igraph)
+importFrom(igraph,max_cliques)
+importFrom(igraph,membership)
 importFrom(igraph,neighbors)
+importFrom(igraph,page_rank)
 importFrom(igraph,simplify)
+importFrom(igraph,strength)
+importFrom(igraph,subgraph.edges)
+importFrom(igraph,subgraph_from_edges)
+importFrom(igraph,transitivity)
 importFrom(igraph,vcount)
+importFrom(igraph,vertex_attr)
 importFrom(igraph,vertex_attr_names)
+importFrom(magrittr,"%>%")
+importFrom(methods,is)
+importFrom(paletteer,scale_color_paletteer_c)
+importFrom(paletteer,scale_fill_paletteer_c)
+importFrom(purrr,map_df)
 importFrom(rlang,.data)
+importFrom(rlang,sym)
+importFrom(scales,pretty_breaks)
+importFrom(stats,median)
+importFrom(stats,runif)
+importFrom(stats,setNames)
+importFrom(tibble,as_tibble)
+importFrom(tibble,rownames_to_column)
+importFrom(tibble,tibble)
 importFrom(tidyr,drop_na)
+importFrom(tidyr,pivot_longer)
+importFrom(tidyr,separate_rows)
+importFrom(utils,head)
+importFrom(utils,setTxtProgressBar)
+importFrom(utils,txtProgressBar)
+importFrom(yulab.utils,str_wrap)

R/PPI_cluster.R

@@ -0,0 +1,174 @@
+#' Compute MCL Clustering for PPI Network
+#'
+#' This function performs Markov Cluster Algorithm (MCL) on an igraph object
+#' and assigns the cluster labels to the nodes.
+#'
+#' @param g An \code{igraph} object.
+#' @param inflation Numeric. The inflation parameter controls the granularity of clusters.
+#'   Common values for PPI: 2.0 - 3.0.
+#'   Larger values (e.g., 3.0) produce smaller, tighter clusters.
+#'   Smaller values (e.g., 1.5) produce larger, coarser clusters.
+#'   Default is 2.5.
+#' @param addLoops Logical; Self-loops with weight 1 are added to each vertex of g when TRUE; Default is TRUE (necessary).
+#' @param ... Additional parameters in function `mcl()`.
+#' @importFrom igraph as_adjacency_matrix V
+#' @importFrom MCL mcl
+#'
+#' @return The input \code{igraph} object with a new vertex attribute \code{mcl_cluster}.
+#' @examples
+#' data(demo_ppi)
+#' library(igraph)
+#' ppi <- run_MCL(demo_ppi, inflation = 2.5)
+#' head(V(ppi)$mcl_cluster)
+#'
+#' @export
+run_MCL <- function(g, inflation = 2.5, addLoops = TRUE, ...) {
+
+  stopifnot(inherits(g, "igraph"))
+  message(sprintf("Running MCL with inflation = %.1f ...", inflation))
+
+  adj_mat <- igraph::as_adjacency_matrix(g, sparse = TRUE)
+
+  ## MCL running
+  mcl_res <- MCL::mcl(x = adj_mat, addLoops = addLoops, inflation = inflation, ...)
+
+  igraph::V(g)$mcl_cluster <- as.factor(mcl_res$Cluster)
+  message(sprintf("Done! Identified %d modules (Iterations: %d).",
+                  mcl_res$K, mcl_res$n.iterations))
+
+  return(g)
+}
+
+
+#' Compute Louvain Clustering for PPI Network
+#'
+#' This function performs Louvain clustering (multi-level modularity optimization)
+#' on an igraph object. It is suitable for detecting larger, macro-scale functional
+#' modules in PPI networks.
+#'
+#' @param g An \code{igraph} object.
+#' @param resolution Numeric. The resolution parameter controls the granularity of clusters. Default is 1.0 (standard modularity).
+#' @param weights Numeric vector or NULL. Edge weights to use for clustering.
+#'   If NULL (default), the function attempts to use the 'weight' or 'score' edge attribute.
+#'   Set to NA to perform unweighted clustering.
+#'
+#' @return The input \code{igraph} object with a new vertex attribute \code{louvain_cluster}.
+#' @importFrom igraph cluster_louvain V edge_attr_names E membership
+#'
+#' @examples
+#' data(demo_ppi)
+#' library(igraph)
+#' ppi <- run_louvain(demo_ppi, resolution = 1)
+#' head(V(ppi)$louvain_cluster)
+#'
+#' @export
+run_louvain <- function(g, resolution = 1.0, weights = NULL) {
+
+  stopifnot(inherits(g, "igraph"))
+
+  # PPI networks often have 'score' or 'weight'. Using them improves accuracy.
+  if (is.null(weights)) {
+    edge_attrs <- igraph::edge_attr_names(g)
+    if ("weight" %in% edge_attrs) {
+      weights <- igraph::E(g)$weight
+      message("Using edge attribute 'weight' for clustering.")
+    } else if ("score" %in% edge_attrs) {
+      weights <- igraph::E(g)$score
+      message("Using edge attribute 'score' for clustering.")
+    } else {
+      weights <- NULL # Unweighted
+      message("No weights found. Running unweighted clustering.")
+    }
+  } else if (identical(weights, NA)) {
+    weights <- NULL # Explicitly unweighted
+  }
+
+  message(sprintf("Running Louvain (Resolution: %.1f)...", resolution))
+
+  # Run Louvain Algorithm
+  louvain_res <- igraph::cluster_louvain(g, weights = weights, resolution = resolution)
+
+  # Assign Cluster Labels
+  igraph::V(g)$louvain_cluster <- as.factor(igraph::membership(louvain_res))
+  num_clusters <- length(unique(igraph::membership(louvain_res)))
+  modularity_score <- max(louvain_res$modularity, na.rm = TRUE)
+
+  message(sprintf("Done! Identified %d modules (Modularity: %.3f).",
+                  num_clusters, modularity_score))
+
+  return(g)
+}
+
+
+#' Score and Rank Network Clusters
+#'
+#' This function evaluates the clusters/modules identified in the graph.
+#' It calculates a score for each cluster based on density and size (MCODE style),
+#' and returns a ranked data frame.
+#'
+#' @param g An \code{igraph} object. The graph must have a vertex attribute containing cluster labels.
+#' @param cluster_attr Character. The name of the vertex attribute that stores cluster labels. Default is louvain_cluster.
+#' @param min_size Integer. Clusters smaller than this size will be ignored. Default is 3.
+#'
+#' @return A data frame containing cluster statistics, ranked by Score.
+#' @importFrom igraph vertex_attr vertex_attr_names induced_subgraph vcount edge_density V ecount
+#' @importFrom utils head
+#' @examples
+#' data(demo_ppi)
+#' ppi <- run_louvain(demo_ppi, resolution = 1)
+#' louvain_score <- Addclusterscore(ppi, cluster_attr = "louvain_cluster", min_size = 3)
+#' head(louvain_score)
+#'
+#' @export
+Addclusterscore <- function(g, cluster_attr = "louvain_cluster", min_size = 3) {
+
+  stopifnot(inherits(g, "igraph"))
+
+  if (!cluster_attr %in% igraph::vertex_attr_names(g)) {
+    stop(paste0("Attribute '", cluster_attr, "' not found in graph. Please run clustering first."))
+  }
+
+  labels <- igraph::vertex_attr(g, cluster_attr)
+  unique_clusters <- unique(labels)
+  unique_clusters <- unique_clusters[!is.na(unique_clusters)]
+  message(paste("Evaluating", length(unique_clusters), "clusters based on attribute:", cluster_attr))
+
+  stats_list <- lapply(unique_clusters, function(cid) {
+    nodes_in_mod <- igraph::V(g)$name[labels == cid]
+    n_nodes <- length(nodes_in_mod)
+
+    if (n_nodes < min_size) return(NULL)
+    subg <- igraph::induced_subgraph(g, nodes_in_mod)
+
+    # Density = E / (N * (N-1) / 2)
+    dens <- igraph::edge_density(subg)
+
+    # Score = Density * N
+    score <- dens * n_nodes
+
+    data.frame(
+      Cluster_ID = as.character(cid),
+      Score      = round(score, 3),
+      Nodes      = n_nodes,
+      Edges      = igraph::ecount(subg),
+      Density    = round(dens, 3),
+      Gene_List  = paste(utils::head(nodes_in_mod, 5), collapse = ", "),
+      Full_Genes = paste(nodes_in_mod, collapse = ","),
+      stringsAsFactors = FALSE
+    )
+  })
+
+  stats_list <- stats_list[!sapply(stats_list, is.null)]
+
+  if (length(stats_list) == 0) {
+    warning("No clusters passed the size filter.")
+    return(data.frame())
+  }
+
+  df_res <- do.call(rbind, stats_list)
+  df_res <- df_res[order(df_res$Score, decreasing = TRUE), ]
+  rownames(df_res) <- NULL
+
+  return(df_res)
+}
+