improved PPI retrieval

sigven · Aug 11, 2020 · 788489b · 788489b
1 parent 1b3ba71
commit 788489b
Show file tree

Hide file tree

Showing 9 changed files with 139 additions and 43 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,7 +1,7 @@
 Package: oncoEnrichR
 Type: Package
 Title: Functional annotation and enrichment analysis of genesets in the context of cancer
-Version: 0.7.0
+Version: 0.7.2
 Author: Sigve Nakken
 Maintainer: Sigve Nakken <[email protected]>
 URL: https://github.com/sigven/oncoEnrichR
@@ -23,5 +23,5 @@ License: MIT
 Encoding: UTF-8
 Remotes: mjkallen/rlogging, jespermaag/gganatogram, molsysbio/speed2
 LazyData: true
-RoxygenNote: 7.1.0
+RoxygenNote: 7.1.1
 Roxygen: list(markdown = TRUE)
diff --git a/R/onco_enrichr.R b/R/onco_enrichr.R
@@ -211,15 +211,17 @@ onco_enrich <- function(query,
   #gtex_atlasassay_groups = c("g32","g9","g29","g10","g28","g44","g33","g50","g37","g38","g42","g35")) {
   stopifnot(is.character(query))
   stopifnot(p_value_adjustment_method %in% c("holm", "hochberg", "hommel", "bonferroni", "BH", "BY", "fdr", "none"))
-  if (length(query) > 1000 | length(query) < 20) {
-    rlogging::message(paste0("ERROR: oncoEnrichR needs minimum 20 query identifiers, and accepts a maximum of 1000. Query contained n = ",length(query), " identifiers"))
+  if (length(query) > 800 | length(query) < 20) {
+    rlogging::message(paste0("ERROR: oncoEnrichR needs minimum 20 query identifiers, and accepts a maximum of 800. Query contained n = ",length(query), " identifiers"))
     return(NULL)
   }
   stopifnot(query_source == "symbol" | query_source == "entrezgene" |
               query_source == "uniprot_acc" | query_source == "ensembl_gene_id")
   stopifnot(ppi_score_threshold > 0 & ppi_score_threshold <= 1000)
   stopifnot(p_value_cutoff_enrichment > 0 & p_value_cutoff_enrichment < 1)
   stopifnot(q_value_cutoff_enrichment > 0 & q_value_cutoff_enrichment < 1)
+  stopifnot(ppi_add_nodes <= 50)
+
 
   qgenes_match <-
     oncoEnrichR::verify_query_genes(query,

diff --git a/R/ppi.R b/R/ppi.R
@@ -68,34 +68,16 @@ get_network_communities <- function(edges = NULL, nodes = NULL){
   return(community_structure)
 }
 
-get_ppi_network <- function(qgenes, ppi_source = "STRING", genedb = NULL,
-                            cancerdrugdb = NULL, settings = NULL){
 
-  stopifnot(!is.null(settings))
-  stopifnot(!is.null(genedb))
-  stopifnot(!is.null(cancerdrugdb))
-  stopifnot(settings$query_type == "interaction_partners" | settings$query_type == "network")
-  oncoEnrichR::validate_db_df(genedb, dbtype = "genedb")
+get_string_network_nodes_edges <- function(qgenes, all_query_nodes = NULL, settings = NULL, genedb = NULL){
 
   query_list <- paste(qgenes, collapse="%0d")
 
-  query_nodes <- data.frame("entrezgene" = qgenes, stringsAsFactors = F) %>%
-    dplyr::distinct() %>%
-    dplyr::left_join(genedb, by = c("entrezgene" = "entrezgene")) %>%
-    dplyr::mutate(id = paste0("s",entrezgene)) %>%
-    dplyr::select(-ensembl_gene_id) %>%
-    dplyr::mutate(query_node = T) %>%
-    dplyr::distinct()
-
-  rlogging::message("STRINGdb: retrieving protein-protein interaction network from (v11)")
-  rlogging::message(paste0("STRINGdb: Settings -  required_score = ",
-                           settings$minimum_score,", add_nodes = ",settings$add_nodes))
-
-  all_edges <- jsonlite::fromJSON(paste0("https://string-db.org/api/json/",
-                                        settings$query_type,"?species=9606&identifiers=",
-                                        query_list,"&required_score=",
-                                        settings$minimum_score,"&add_nodes=",settings$add_nodes)) %>%
-    dplyr::left_join(dplyr::select(genedb,entrezgene,symbol),by=c("preferredName_A" = "symbol")) %>%
+  edges <- jsonlite::fromJSON(paste0("https://string-db.org/api/json/",
+                                         settings$query_type, "?species=9606&identifiers=",
+                                         query_list, "&required_score=",
+                                         settings$minimum_score, "&add_nodes=", settings$add_nodes)) %>%
+    dplyr::left_join(dplyr::select(genedb,entrezgene,symbol), by = c("preferredName_A" = "symbol")) %>%
     dplyr::filter(!is.na(entrezgene)) %>%
     dplyr::rename(entrezgene_a = entrezgene) %>%
     dplyr::mutate(entrezgene_a = as.character(entrezgene_a)) %>%
@@ -104,7 +86,7 @@ get_ppi_network <- function(qgenes, ppi_source = "STRING", genedb = NULL,
                                    cancer_driver, tcga_driver),by=c("entrezgene_a" = "entrezgene")) %>%
     dplyr::rename(oncogene_A = oncogene, tsgene_A = tumor_suppressor,
                   cdriver_A = cancer_driver, tcga_driver_A = tcga_driver) %>%
-    dplyr::left_join(dplyr::select(genedb,entrezgene,symbol),by=c("preferredName_B" = "symbol")) %>%
+    dplyr::left_join(dplyr::select(genedb,entrezgene,symbol), by = c("preferredName_B" = "symbol")) %>%
     dplyr::filter(!is.na(entrezgene)) %>%
     dplyr::rename(entrezgene_b = entrezgene) %>%
     dplyr::mutate(entrezgene_b = as.character(entrezgene_b)) %>%
@@ -114,29 +96,141 @@ get_ppi_network <- function(qgenes, ppi_source = "STRING", genedb = NULL,
     dplyr::rename(oncogene_B = oncogene, tsgene_B = tumor_suppressor, cdriver_B = cancer_driver,
                   tcga_driver_B = tcga_driver) %>%
     dplyr::mutate(interaction_symbol = paste0(preferredName_A,"_",preferredName_B)) %>%
-    dplyr::left_join(dplyr::select(query_nodes, symbol, query_node), by = c("preferredName_A" = "symbol")) %>%
+    dplyr::left_join(dplyr::select(all_query_nodes, symbol, query_node), by = c("preferredName_A" = "symbol")) %>%
     dplyr::rename(query_node_A = query_node) %>%
-    dplyr::left_join(dplyr::select(query_nodes, symbol, query_node), by = c("preferredName_B" = "symbol")) %>%
+    dplyr::left_join(dplyr::select(all_query_nodes, symbol, query_node), by = c("preferredName_B" = "symbol")) %>%
     dplyr::rename(query_node_B = query_node) %>%
     dplyr::mutate(weight = score) %>%
     dplyr::distinct() %>%
     dplyr::select(-c(ncbiTaxonId,stringId_A,stringId_B))
 
-  network_nodes <- data.frame("symbol" = unique(c(all_edges$preferredName_A, all_edges$preferredName_B)),
+  nodes <- data.frame("symbol" = unique(c(edges$preferredName_A, edges$preferredName_B)),
                               stringsAsFactors = F) %>%
     dplyr::distinct() %>%
     dplyr::left_join(genedb, by = c("symbol" = "symbol")) %>%
     dplyr::select(-ensembl_gene_id) %>%
     dplyr::filter(!is.na(entrezgene)) %>%
-    dplyr::left_join(dplyr::select(query_nodes, symbol, query_node), by = c("symbol")) %>%
+    dplyr::left_join(dplyr::select(all_query_nodes, symbol, query_node), by = c("symbol")) %>%
     dplyr::mutate(id = paste0("s",entrezgene)) %>%
-    dplyr::distinct()
-
-  all_nodes <- dplyr::bind_rows(query_nodes, network_nodes) %>%
+    dplyr::distinct() %>%
+    dplyr::bind_rows(all_query_nodes) %>%
     dplyr::distinct() %>%
     dplyr::mutate(query_node = dplyr::if_else(is.na(query_node), FALSE, as.logical(query_node)))
 
 
+  #all_nodes <- dplyr::bind_rows(query_nodes, network_nodes) %>%
+    #dplyr::distinct() %>%
+    #dplyr::mutate(query_node = dplyr::if_else(is.na(query_node), FALSE, as.logical(query_node)))
+
+
+  network <- list()
+  network$edges <- edges
+  network$nodes <- nodes
+
+  return(network)
+
+
+}
+
+get_ppi_network <- function(qgenes, ppi_source = "STRING", genedb = NULL,
+                            cancerdrugdb = NULL, settings = NULL){
+
+  stopifnot(!is.null(settings))
+  stopifnot(!is.null(genedb))
+  stopifnot(!is.null(cancerdrugdb))
+  stopifnot(settings$query_type == "interaction_partners" | settings$query_type == "network")
+  oncoEnrichR::validate_db_df(genedb, dbtype = "genedb")
+
+  query_nodes <- data.frame("entrezgene" = qgenes, stringsAsFactors = F) %>%
+    dplyr::distinct() %>%
+    dplyr::left_join(genedb, by = c("entrezgene" = "entrezgene")) %>%
+    dplyr::mutate(id = paste0("s",entrezgene)) %>%
+    dplyr::select(-ensembl_gene_id) %>%
+    dplyr::mutate(query_node = T) %>%
+    dplyr::distinct()
+
+  rlogging::message("STRINGdb: retrieving protein-protein interaction network from (v11)")
+  rlogging::message(paste0("STRINGdb: Settings -  required_score = ",
+                           settings$minimum_score,", add_nodes = ",settings$add_nodes))
+
+  all_edges <- data.frame()
+  all_nodes <- data.frame()
+
+  if(length(qgenes) > 200){
+    i <- 1
+    omnipathdb <- data.frame()
+    while(i <= length(qgenes)){
+      qgenes_set <- qgenes[i:min(length(qgenes),i + 199)]
+
+      ppi_network_data <- get_string_network_nodes_edges(qgenes = qgenes_set,
+                                                         all_query_nodes = query_nodes,
+                                                         settings = settings,
+                                                         genedb = genedb)
+
+      all_edges <- all_edges %>% dplyr::bind_rows(ppi_network_data$edges) %>% dplyr::distinct()
+      all_nodes <- all_nodes %>% dplyr::bind_rows(ppi_network_data$nodes) %>% dplyr::distinct()
+      i <- i + 199
+    }
+
+  }else{
+    ppi_network_data <- get_string_network_nodes_edges(qgenes = qgenes,
+                                                       all_query_nodes = query_nodes,
+                                                       settings = settings,
+                                                       genedb = genedb)
+    all_edges <- all_edges %>% dplyr::bind_rows(ppi_network_data$edges) %>% dplyr::distinct()
+    all_nodes <- all_nodes %>% dplyr::bind_rows(ppi_network_data$nodes) %>% dplyr::distinct()
+  }
+
+
+
+  # query_list <- paste(qgenes, collapse="%0d")
+  #
+  # all_edges <- jsonlite::fromJSON(paste0("https://string-db.org/api/json/",
+  #                                       settings$query_type, "?species=9606&identifiers=",
+  #                                       query_list, "&required_score=",
+  #                                       settings$minimum_score, "&add_nodes=", settings$add_nodes)) %>%
+  #   dplyr::left_join(dplyr::select(genedb,entrezgene,symbol), by = c("preferredName_A" = "symbol")) %>%
+  #   dplyr::filter(!is.na(entrezgene)) %>%
+  #   dplyr::rename(entrezgene_a = entrezgene) %>%
+  #   dplyr::mutate(entrezgene_a = as.character(entrezgene_a)) %>%
+  #   dplyr::mutate(from = paste0("s",entrezgene_a)) %>%
+  #   dplyr::left_join(dplyr::select(genedb, entrezgene, oncogene, tumor_suppressor,
+  #                                  cancer_driver, tcga_driver),by=c("entrezgene_a" = "entrezgene")) %>%
+  #   dplyr::rename(oncogene_A = oncogene, tsgene_A = tumor_suppressor,
+  #                 cdriver_A = cancer_driver, tcga_driver_A = tcga_driver) %>%
+  #   dplyr::left_join(dplyr::select(genedb,entrezgene,symbol), by = c("preferredName_B" = "symbol")) %>%
+  #   dplyr::filter(!is.na(entrezgene)) %>%
+  #   dplyr::rename(entrezgene_b = entrezgene) %>%
+  #   dplyr::mutate(entrezgene_b = as.character(entrezgene_b)) %>%
+  #   dplyr::mutate(to = paste0("s",entrezgene_b)) %>%
+  #   dplyr::left_join(dplyr::select(genedb, entrezgene, oncogene, tumor_suppressor,
+  #                                  cancer_driver, tcga_driver), by = c("entrezgene_a" = "entrezgene")) %>%
+  #   dplyr::rename(oncogene_B = oncogene, tsgene_B = tumor_suppressor, cdriver_B = cancer_driver,
+  #                 tcga_driver_B = tcga_driver) %>%
+  #   dplyr::mutate(interaction_symbol = paste0(preferredName_A,"_",preferredName_B)) %>%
+  #   dplyr::left_join(dplyr::select(query_nodes, symbol, query_node), by = c("preferredName_A" = "symbol")) %>%
+  #   dplyr::rename(query_node_A = query_node) %>%
+  #   dplyr::left_join(dplyr::select(query_nodes, symbol, query_node), by = c("preferredName_B" = "symbol")) %>%
+  #   dplyr::rename(query_node_B = query_node) %>%
+  #   dplyr::mutate(weight = score) %>%
+  #   dplyr::distinct() %>%
+  #   dplyr::select(-c(ncbiTaxonId,stringId_A,stringId_B))
+  #
+  # network_nodes <- data.frame("symbol" = unique(c(all_edges$preferredName_A, all_edges$preferredName_B)),
+  #                             stringsAsFactors = F) %>%
+  #   dplyr::distinct() %>%
+  #   dplyr::left_join(genedb, by = c("symbol" = "symbol")) %>%
+  #   dplyr::select(-ensembl_gene_id) %>%
+  #   dplyr::filter(!is.na(entrezgene)) %>%
+  #   dplyr::left_join(dplyr::select(query_nodes, symbol, query_node), by = c("symbol")) %>%
+  #   dplyr::mutate(id = paste0("s",entrezgene)) %>%
+  #   dplyr::distinct()
+  #
+  # all_nodes <- dplyr::bind_rows(query_nodes, network_nodes) %>%
+  #   dplyr::distinct() %>%
+  #   dplyr::mutate(query_node = dplyr::if_else(is.na(query_node), FALSE, as.logical(query_node)))
+
+
   all_nodes <- all_nodes %>%
     dplyr::mutate(shape = dplyr::if_else(query_node == T,settings$visnetwork_shape, as.character("box")))
   all_nodes$shadow <- settings$visnetwork_shadow
@@ -151,7 +245,6 @@ get_ppi_network <- function(qgenes, ppi_source = "STRING", genedb = NULL,
   all_nodes$size <- 25
   all_nodes <- all_nodes %>%
     dplyr::mutate(color.background  = dplyr::if_else(query_node == T, "lightblue", "mistyrose")) %>%
-    #dplyr::mutate(tumor_suppressor = )
     dplyr::mutate(color.background = dplyr::if_else(tumor_suppressor == T & oncogene == F,
                                                     "firebrick", as.character(color.background),
                                                     as.character(color.background))) %>%

diff --git a/data/alias2primary.rda b/data/alias2primary.rda
diff --git a/data/otdb.rda b/data/otdb.rda
diff --git a/data/release_notes.rda b/data/release_notes.rda
diff --git a/data/wikipathwaydb.rda b/data/wikipathwaydb.rda
diff --git a/man/onco_enrich.Rd b/man/onco_enrich.Rd
diff --git a/oncoEnrichR_0.7.1.pdf → oncoEnrichR_0.7.2.pdf b/oncoEnrichR_0.7.1.pdf → oncoEnrichR_0.7.2.pdf