R/ppi.R
In sigven/oncoEnrichR: Cancer-dedicated gene set interpretation
Defines functions
get_ppi_network
get_string_network_nodes_edges
get_biogrid_network_nodes_edges
get_network_communities
get_network_hubs
get_network_hubs <- function(edges = NULL,
nodes = NULL,
genedb = NULL) {
stopifnot(!is.null(edges) & !is.null(nodes))
validate_db_df(nodes, dbtype = "ppi_nodes")
validate_db_df(edges, dbtype = "ppi_edges")
validate_db_df(genedb, dbtype = "genedb")
edges <- edges |>
dplyr::filter(
!is.na(.data$symbol_A) &
!is.na(.data$symbol_B)) |>
dplyr::select(
c("symbol_A", "symbol_B", "from",
"to", "weight"))
d <- igraph::graph_from_data_frame(d = edges, directed = F)
## hub score (Kleinberg"s hub centrality)
hscore <- igraph::hub_score(d)
hub_scores <- data.frame(
symbol = names(sort(hscore$vector,decreasing = T)),
hub_score = round(sort(hscore$vector,decreasing = T), digits = 3),
stringsAsFactors = F) |>
dplyr::left_join(
dplyr::select(
genedb, c("symbol", "name")),
by = c("symbol"), relationship = "many-to-many") |>
dplyr::left_join(
dplyr::select(
nodes, c("symbol", "query_node")),
by = c("symbol"), relationship = "many-to-many") |>
dplyr::filter(.data$query_node == T) |>
dplyr::select(c("symbol", "name", "hub_score")) |>
dplyr::distinct()
#closeness_score <- igraph::closeness(d, mode="all")
#degree.cent <- centr_degree(d, mode = "all")
return(hub_scores)
}
get_network_communities <- function(edges = NULL, nodes = NULL) {
stopifnot(!is.null(edges) & !is.null(nodes))
validate_db_df(nodes, dbtype = "ppi_nodes")
validate_db_df(edges, dbtype = "ppi_edges")
edges <- edges |>
dplyr::filter(
!is.na(.data$symbol_A) &
!is.na(.data$symbol_B)) |>
dplyr::select(
c("symbol_A","symbol_B",
"from","to","weight"))
d <- igraph::graph_from_data_frame(d = edges, directed = F)
## communities, fast greedy modularity optimization algorithm
## for finding community structure,
cties <- igraph::fastgreedy.community(d)
edges_communities <- data.frame()
if (length(cties) > 0) {
n <- 1
while(n <= length(cties)) {
members_community_n <- cties[[n]]
community <- dplyr::filter(
edges, .data$symbol_A %in% members_community_n &
.data$symbol_B %in% members_community_n) |>
dplyr::mutate(community = n)
edges_communities <- edges_communities |>
dplyr::bind_rows(community)
n <- n + 1
}
}
nodes_communities <- data.frame(
"id" = unique(c(unique(edges_communities$from),
unique(edges_communities$to))),
stringsAsFactors = F) |>
dplyr::inner_join(nodes, by = c("id"), relationship = "many-to-many") |>
dplyr::distinct()
community_structure <- list()
community_structure[["edges"]] <- edges_communities
community_structure[["nodes"]] <- nodes_communities
return(community_structure)
}
get_biogrid_network_nodes_edges <-
function(all_query_nodes = NULL,
add_nodes = 50,
minimum_evidence_items = 2,
show_isolated_nodes = FALSE,
genedb = NULL,
biogrid = NULL) {
stopifnot(!is.null(all_query_nodes))
stopifnot(is.data.frame(all_query_nodes))
stopifnot(NROW(all_query_nodes) > 1)
ppi_edges <- data.frame()
## match edges (interactions) in BioGRID entrezgene_B
edges_part1 <- biogrid |>
dplyr::inner_join(
dplyr::select(all_query_nodes, c("entrezgene","symbol")),
by = c("entrezgene_B" = "entrezgene"),
relationship = "many-to-many"
) |>
dplyr::rename(symbol_B = "symbol") |>
dplyr::left_join(
dplyr::select(
genedb, c("entrezgene", "symbol")),
by = c("entrezgene_A" = "entrezgene"),
relationship = "many-to-many"
) |>
dplyr::rename(symbol_A = "symbol") |>
dplyr::arrange(
.data$symbol_A,
.data$symbol_B,
dplyr::desc(.data$pmid)) |>
dplyr::mutate(
evidence = paste(
.data$method,
.data$throughput,
.data$pmid, sep ="|"
)
)
## match edges (interactions) in BioGRID against entrezgene_B
edges_part2 <- biogrid |>
dplyr::inner_join(
dplyr::select(all_query_nodes, c("entrezgene","symbol")),
by = c("entrezgene_A" = "entrezgene"),
relationship = "many-to-many"
) |>
dplyr::rename(symbol_A = "symbol") |>
dplyr::left_join(
dplyr::select(
genedb, c("entrezgene", "symbol")),
by = c("entrezgene_B" = "entrezgene"),
relationship = "many-to-many"
) |>
dplyr::rename(symbol_B = "symbol") |>
dplyr::arrange(.data$symbol_A,
.data$symbol_B,
dplyr::desc(.data$pmid)) |>
dplyr::mutate(
evidence = paste(
.data$method,
.data$throughput,
.data$pmid, sep ="|"
)
)
all_edges <- as.data.frame(
dplyr::bind_rows(
edges_part1,
edges_part2) |>
dplyr::distinct() |>
dplyr::filter(.data$symbol_A != .data$symbol_B) |>
dplyr::arrange(.data$symbol_A, .data$symbol_B) |>
dplyr::group_by(
.data$entrezgene_A,
.data$entrezgene_B,
.data$symbol_A,
.data$symbol_B) |>
dplyr::summarise(
evidence = paste(
unique(.data$evidence), collapse = ","
),
num_evidence_items = dplyr::n(),
.groups = "drop") |>
## only keep interactions with a minimum number
## of evidence items
dplyr::filter(
.data$num_evidence_items >= minimum_evidence_items) |>
dplyr::mutate(
querynode_A = dplyr::if_else(
.data$entrezgene_A %in% all_query_nodes$entrezgene,
TRUE,
FALSE
),
querynode_B = dplyr::if_else(
.data$entrezgene_B %in% all_query_nodes$entrezgene,
TRUE,
FALSE
)
)
)
## network - query nodes only
ppi_edges <- all_edges |>
dplyr::filter(
.data$querynode_A == T &
.data$querynode_B == T)
if (add_nodes > 0) {
## attach non-query nodes with most interactions with query nodes
edges_single_node_inqueryset <- all_edges |>
dplyr::filter(.data$querynode_A == F |
.data$querynode_B == F)
if (NROW(edges_single_node_inqueryset) > 0) {
set1 <- edges_single_node_inqueryset |>
dplyr::filter(.data$querynode_A == F)
if (NROW(set1) > 0) {
set1 <- set1 |>
dplyr::select(c("entrezgene_A", "entrezgene_B")) |>
dplyr::distinct() |>
dplyr::group_by(.data$entrezgene_A) |>
dplyr::summarise(num_interactions = dplyr::n()) |>
dplyr::arrange(dplyr::desc(.data$num_interactions)) |>
dplyr::rename(entrezgene = "entrezgene_A")
}
set2 <- edges_single_node_inqueryset |>
dplyr::filter(.data$querynode_B == F)
if (NROW(set2) > 0) {
set2 <- set2 |>
dplyr::select(c("entrezgene_A", "entrezgene_B")) |>
dplyr::distinct() |>
dplyr::group_by(.data$entrezgene_B) |>
dplyr::summarise(num_interactions = dplyr::n()) |>
dplyr::arrange(dplyr::desc(.data$num_interactions)) |>
dplyr::rename(entrezgene = "entrezgene_B")
}
if (NROW(set1) > 0 | NROW(set2) > 0) {
all_non_querynodes_ranked <-
dplyr::bind_rows(
set1, set2) |>
dplyr::arrange(
dplyr::desc(.data$num_interactions)) |>
utils::head(add_nodes)
edges_added <-
edges_single_node_inqueryset |>
dplyr::filter(
.data$entrezgene_A %in% all_non_querynodes_ranked$entrezgene |
.data$entrezgene_B %in% all_non_querynodes_ranked$entrezgene
)
if (NROW(edges_added) > 0) {
ppi_edges <- dplyr::bind_rows(
ppi_edges, edges_added
)
}
}
}
}
network <- NULL
## ignore proto-oncogenes/tumor-suppressors with weak
## support/confidence in the protein-protein interaction network
genedb <- genedb |>
dplyr::mutate(oncogene = dplyr::if_else(
.data$oncogene_confidence_level == "MODERATE",
FALSE,
as.logical(.data$oncogene)
)) |>
dplyr::mutate(tumor_suppressor = dplyr::if_else(
.data$tsg_confidence_level == "MODERATE",
FALSE,
as.logical(.data$tumor_suppressor)
))
if(NROW(ppi_edges) > 0){
edges <- ppi_edges |>
dplyr::mutate(from = paste0("s", .data$entrezgene_A)) |>
dplyr::left_join(
dplyr::select(
genedb,
c("entrezgene","oncogene",
"tumor_suppressor",
"cancer_driver")),
by = c("entrezgene_A" = "entrezgene"), relationship = "many-to-many") |>
dplyr::rename(oncogene_A = "oncogene",
tsgene_A = "tumor_suppressor",
cdriver_A = "cancer_driver") |>
dplyr::mutate(to = paste0("s", .data$entrezgene_B)) |>
dplyr::left_join(
dplyr::select(genedb,
c("entrezgene", "oncogene",
"tumor_suppressor","cancer_driver")),
by = c("entrezgene_B" = "entrezgene"), relationship = "many-to-many") |>
dplyr::rename(oncogene_B = "oncogene",
tsgene_B = "tumor_suppressor",
cdriver_B = "cancer_driver") |>
dplyr::mutate(interaction_symbol =
paste0(.data$symbol_A, "_",
.data$symbol_B)) |>
dplyr::mutate(weight = dplyr::if_else(
.data$num_evidence_items >= 5,
as.numeric(2),
as.numeric(1)
)) |>
#dplyr::mutate(weight = .data$num_evidence_items) |>
dplyr::mutate(title = paste0("Number of evidence items:",
.data$num_evidence_items)) |>
dplyr::distinct()
nodes <- data.frame(
"entrezgene" = unique(c(edges$entrezgene_A,
edges$entrezgene_B)),
stringsAsFactors = F) |>
dplyr::distinct() |>
dplyr::left_join(dplyr::select(
genedb,
c("symbol","entrezgene",
"tumor_suppressor","oncogene",
"cancer_driver","genename",
"targeted_cancer_drugs_ep",
"targeted_cancer_drugs_lp")),
by = "entrezgene", relationship = "many-to-many") |>
dplyr::filter(!is.na(.data$entrezgene)) |>
dplyr::left_join(
dplyr::select(all_query_nodes,
c("entrezgene","query_node")),
by = "entrezgene", relationship = "many-to-many") |>
dplyr::mutate(id = paste0("s", .data$entrezgene)) |>
dplyr::distinct() |>
dplyr::mutate(query_node = dplyr::if_else(
is.na(.data$query_node),
FALSE,
as.logical(.data$query_node)))
if(show_isolated_nodes == T){
nodes <- nodes |>
dplyr::bind_rows(all_query_nodes) |>
dplyr::distinct() |>
dplyr::mutate(query_node = dplyr::if_else(
is.na(.data$query_node),
FALSE,
as.logical(.data$query_node)))
}
network <- list()
network$edges <- edges
network$nodes <- nodes
}
return(network)
}
get_string_network_nodes_edges <-
function(qgenes,
all_query_nodes = NULL,
query_type = "network",
minimum_score = 0.9,
add_nodes = 50,
show_isolated_nodes = FALSE,
network_type = "physical",
genedb = NULL) {
query_list <- paste(qgenes, collapse = "%0d")
edges <- jsonlite::fromJSON(
paste0("https://string-db.org/api/json/",
query_type,
"?species=9606&identifiers=", query_list,
"&required_score=", as.integer(minimum_score * 1000),
"&network_type=", network_type,
"&add_nodes=", add_nodes))
## ignore proto-oncogenes/tumor-suppressors with weak
## support/confidence in the protein-protein interaction network
genedb <- genedb |>
dplyr::mutate(oncogene = dplyr::if_else(
.data$oncogene_confidence_level == "MODERATE",
FALSE,
as.logical(.data$oncogene)
)) |>
dplyr::mutate(tumor_suppressor = dplyr::if_else(
.data$tsg_confidence_level == "MODERATE",
FALSE,
as.logical(.data$tumor_suppressor)
))
if (NROW(edges) > 0) {
edges <- edges |>
dplyr::rename(symbol_A = "preferredName_A",
symbol_B = "preferredName_B") |>
dplyr::left_join(
dplyr::select(genedb, c("entrezgene", "symbol")),
by = c("symbol_A" = "symbol"), relationship = "many-to-many") |>
dplyr::filter(!is.na(.data$entrezgene)) |>
dplyr::rename(entrezgene_A = "entrezgene") |>
dplyr::mutate(from = paste0("s", .data$entrezgene_A)) |>
dplyr::left_join(
dplyr::select(
genedb,
c("entrezgene","oncogene",
"tumor_suppressor",
"cancer_driver")),
by = c("entrezgene_A" = "entrezgene"), relationship = "many-to-many") |>
dplyr::rename(oncogene_A = "oncogene",
tsgene_A = "tumor_suppressor",
cdriver_A = "cancer_driver") |>
dplyr::left_join(dplyr::select(genedb, c("entrezgene", "symbol")),
by = c("symbol_B" = "symbol"), relationship = "many-to-many") |>
dplyr::filter(!is.na(.data$entrezgene)) |>
dplyr::rename(entrezgene_B = "entrezgene") |>
dplyr::mutate(to = paste0("s", .data$entrezgene_B)) |>
dplyr::left_join(
dplyr::select(genedb,
c("entrezgene", "oncogene",
"tumor_suppressor","cancer_driver")),
by = c("entrezgene_B" = "entrezgene"), relationship = "many-to-many") |>
dplyr::rename(oncogene_B = "oncogene",
tsgene_B = "tumor_suppressor",
cdriver_B = "cancer_driver") |>
dplyr::mutate(interaction_symbol =
paste0(.data$symbol_A, "_",
.data$symbol_B)) |>
dplyr::left_join(
dplyr::select(all_query_nodes, c("symbol","query_node")),
by = c("symbol_A" = "symbol"), relationship = "many-to-many") |>
dplyr::rename(querynode_A = "query_node") |>
dplyr::left_join(
dplyr::select(all_query_nodes, c("symbol","query_node")),
by = c("symbol_B" = "symbol"), relationship = "many-to-many") |>
dplyr::rename(querynode_B = "query_node") |>
dplyr::mutate(
querynode_A = dplyr::if_else(
is.na(.data$querynode_A),
FALSE,
as.logical(.data$querynode_A)
),
querynode_B = dplyr::if_else(
is.na(.data$querynode_B),
FALSE,
as.logical(.data$querynode_B)
)
) |>
dplyr::mutate(weight = .data$score) |>
dplyr::mutate(title = paste0("Interaction score:", .data$score)) |>
dplyr::distinct() |>
dplyr::select(-c("ncbiTaxonId", "stringId_A", "stringId_B"))
nodes <- data.frame(
"symbol" = unique(c(edges$symbol_A, edges$symbol_B)),
stringsAsFactors = F) |>
dplyr::distinct() |>
dplyr::left_join(dplyr::select(
genedb,
c("symbol","entrezgene",
"tumor_suppressor","oncogene",
"cancer_driver","genename",
"targeted_cancer_drugs_ep",
"targeted_cancer_drugs_lp")),
by = "symbol", relationship = "many-to-many") |>
dplyr::filter(!is.na(.data$entrezgene)) |>
dplyr::left_join(
dplyr::select(all_query_nodes,
c("symbol","query_node")),
by = "symbol", relationship = "many-to-many") |>
dplyr::mutate(id = paste0("s", .data$entrezgene)) |>
dplyr::distinct() |>
dplyr::mutate(query_node = dplyr::if_else(
is.na(.data$query_node),
FALSE,
as.logical(.data$query_node)))
if(show_isolated_nodes == T){
nodes <- nodes |>
dplyr::bind_rows(all_query_nodes) |>
dplyr::distinct() |>
dplyr::mutate(query_node = dplyr::if_else(
is.na(.data$query_node),
FALSE,
as.logical(.data$query_node)))
}
network <- list()
network$edges <- edges
network$nodes <- nodes
}
else{
network <- NULL
}
return(network)
}
get_ppi_network <- function(qgenes = NULL,
ppi_source = "string",
genedb = NULL,
cancerdrugdb = NULL,
biogrid = NULL,
settings = NULL,
ppi_source_release = NA) {
lgr::lgr$appenders$console$set_layout(
lgr::LayoutFormat$new(timestamp_fmt = "%Y-%m-%d %T"))
stopifnot(!is.null(qgenes))
stopifnot(!is.null(settings))
stopifnot(!is.null(genedb))
validate_db_df(genedb, dbtype = "genedb")
stopifnot(is.integer(qgenes))
stopifnot(!is.null(cancerdrugdb))
stopifnot(!is.null(biogrid))
validate_db_df(biogrid, dbtype = "biogrid")
stopifnot(!is.null(cancerdrugdb[['network']]))
stopifnot(ppi_source == "string" | ppi_source == "biogrid")
if (ppi_source == "string") {
stopifnot(identical(
names(settings[[ppi_source]]),
c("minimum_score",
"visnetwork_shape",
"visnetwork_shadow",
"show_drugs",
"add_nodes",
"query_type",
"network_type",
"show_isolated_nodes"))
)
stopifnot(settings[[ppi_source]]$query_type == "interaction_partners" |
settings[[ppi_source]]$query_type == "network")
stopifnot(settings[[ppi_source]]$network_type == "physical" |
settings[[ppi_source]]$network_type == "functional")
}
if (ppi_source == "biogrid") {
stopifnot(identical(
names(settings[[ppi_source]]),
c("minimum_evidence",
"visnetwork_shape",
"visnetwork_shadow",
"show_drugs",
"add_nodes",
"show_isolated_nodes"))
)
}
query_nodes <- data.frame("entrezgene" = qgenes, stringsAsFactors = F) |>
dplyr::distinct() |>
dplyr::left_join(dplyr::select(
genedb,
c("symbol","entrezgene","tumor_suppressor",
"cancer_driver", "oncogene", "genename",
"targeted_cancer_drugs_ep",
"targeted_cancer_drugs_lp")),
by = c("entrezgene" = "entrezgene"), relationship = "many-to-many") |>
dplyr::mutate(id = paste0("s", .data$entrezgene)) |>
dplyr::mutate(query_node = T) |>
dplyr::distinct()
## Occasionally, some entrezgene identifiers are ambiguous (wrt. gene symbols)
## if this is the case, pick random entry from duplicate records, so that
## all entrezgene identifiers point to a unique row
if (nrow(query_nodes) > length(qgenes)) {
duplicate_records <- as.data.frame(
query_nodes |>
dplyr::group_by(.data$entrezgene) |>
dplyr::summarise(n = dplyr::n(),
.groups = "drop") |>
dplyr::filter(.data$n > 1)
)
lgr::lgr$info( paste0(
"Resolving ambiguous Entrez gene identifiers: ",
paste(duplicate_records$entrezgene, collapse = ", ")))
query_nodes_clean <-
as.data.frame(
query_nodes |>
dplyr::group_by(.data$entrezgene) |>
dplyr::summarise(n = dplyr::n(),
.groups = "drop") |>
dplyr::filter(.data$n == 1)
)
for (i in 1:nrow(duplicate_records)) {
entrezgene_id = duplicate_records[i,]$entrezgene
sample_row <- dplyr::sample_n(
dplyr::filter(query_nodes, .data$entrezgene == entrezgene_id),
1
)
query_nodes_clean <- query_nodes_clean |>
dplyr::bind_rows(sample_row)
i <- i + 1
}
if (length(qgenes) == nrow(query_nodes_clean)) {
query_nodes <- query_nodes_clean
}
}
ppi_source_verbose <- "BioGRID"
if (ppi_source == "string") {
ppi_source_verbose <- "STRING"
lgr::lgr$info( paste0(
ppi_source_verbose,
": retrieving protein-protein interaction network from API (",
ppi_source_release,")"))
lgr::lgr$info( paste0(
ppi_source_verbose,
": Settings - required minimum score = ",
settings[[ppi_source]]$minimum_score,
", add_nodes = ",settings[[ppi_source]]$add_nodes,
", show_isolated_nodes = ",settings[[ppi_source]]$show_isolated_nodes,
", network_type = ", settings[[ppi_source]]$network_type))
all_edges <- data.frame()
all_nodes <- data.frame()
if (length(qgenes) > 200) {
i <- 1
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,
query_type = settings[[ppi_source]]$query_type,
network_type = settings[[ppi_source]]$network_type,
minimum_score = settings[[ppi_source]]$minimum_score,
add_nodes = settings[[ppi_source]]$add_nodes,
show_isolated_nodes = settings[[ppi_source]]$show_isolated_nodes,
genedb = genedb)
all_edges <- all_edges |>
dplyr::bind_rows(ppi_network_data$edges) |>
dplyr::distinct()
all_nodes <- all_nodes |>
dplyr::bind_rows(ppi_network_data$nodes) |>
dplyr::distinct()
i <- i + 199
}
} else {
ppi_network_data <- get_string_network_nodes_edges(
qgenes = qgenes,
all_query_nodes = query_nodes,
query_type = settings[[ppi_source]]$query_type,
network_type = settings[[ppi_source]]$network_type,
minimum_score = settings[[ppi_source]]$minimum_score,
show_isolated_nodes = settings[[ppi_source]]$show_isolated_nodes,
add_nodes = settings[[ppi_source]]$add_nodes,
genedb = genedb)
if (!is.null(ppi_network_data)) {
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()
}
}
}
if (ppi_source == "biogrid") {
lgr::lgr$info(paste0(
ppi_source_verbose,
": retrieving physical protein-protein interaction network (",
ppi_source_release,")"))
lgr::lgr$info( paste0(
ppi_source_verbose,
": Settings - minimum evidence = ",
settings[[ppi_source]]$minimum_evidence,
", show_isolated_nodes = ",settings[[ppi_source]]$show_isolated_nodes,
", add_nodes = ",settings[[ppi_source]]$add_nodes))
all_edges <- data.frame()
all_nodes <- data.frame()
ppi_network_data <- get_biogrid_network_nodes_edges(
all_query_nodes = query_nodes,
minimum_evidence_items = settings[[ppi_source]]$minimum_evidence,
add_nodes = settings[[ppi_source]]$add_nodes,
show_isolated_nodes = settings[[ppi_source]]$show_isolated_nodes,
biogrid = biogrid,
genedb = genedb)
if (!is.null(ppi_network_data)) {
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()
}
}
if (nrow(all_edges) > 0 & nrow(all_nodes) > 0) {
all_nodes <- all_nodes |>
dplyr::mutate(shape = dplyr::if_else(
.data$query_node == T,
settings[[ppi_source]]$visnetwork_shape,
as.character("box")))
all_nodes$shadow <- settings[[ppi_source]]$visnetwork_shadow
all_nodes$title <- unlist(
lapply(stringr::str_match_all(
all_nodes$genename,">.+<"), paste, collapse = ","))
all_nodes$title <- stringr::str_replace_all(
all_nodes$title,">|<", "")
all_nodes$label <- all_nodes$symbol # Node label
all_edges$width <- all_edges$weight
all_nodes$gene_category <- "protein_coding"
all_nodes$size <- 25
all_nodes <- all_nodes |>
dplyr::mutate(color.background = dplyr::if_else(
.data$query_node == T, "lightblue", "mistyrose")) |>
dplyr::mutate(color.background = dplyr::if_else(
.data$tumor_suppressor == T & .data$oncogene == F,
"firebrick", as.character(.data$color.background),
as.character(.data$color.background))) |>
dplyr::mutate(color.background = dplyr::if_else(
.data$oncogene == T & .data$tumor_suppressor == F,
"darkolivegreen", as.character(.data$color.background),
as.character(.data$color.background))) |>
dplyr::mutate(color.background = dplyr::if_else(
.data$oncogene == T & .data$tumor_suppressor == T,
"black", as.character(.data$color.background),
as.character(.data$color.background))) |>
dplyr::mutate(color.border = "black",
color.highlight.background = "orange",
color.highlight.border = "darkred",
font.color = "white") |>
dplyr::mutate(font.color = dplyr::if_else(
.data$query_node == T | .data$color.background == "mistyrose",
"black", as.character(.data$font.color),
as.character(.data$font.color)))
nodes_with_drugs <- all_nodes
edges_with_drugs <- all_edges
if (settings[[ppi_source]]$show_drugs == T) {
drug_target_ids <-
dplyr::inner_join(
dplyr::select(all_nodes, c("id")),
dplyr::select(cancerdrugdb[['network']]$edges, c("from")),
by = c("id" = "from"), relationship = "many-to-many") |>
dplyr::distinct()
if (nrow(drug_target_ids) > 0) {
drug_edges <- dplyr::inner_join(
cancerdrugdb[['network']]$edges,
drug_target_ids, by = c("from" = "id"))
drug_nodes <- drug_edges |>
dplyr::select(-c("from")) |>
dplyr::inner_join(cancerdrugdb[['network']]$nodes,
by = c("to" = "id"), relationship = "many-to-many") |>
dplyr::rename(id = "to") |>
dplyr::distinct()
nodes_with_drugs <- dplyr::bind_rows(all_nodes, drug_nodes)
edges_with_drugs <- dplyr::bind_rows(all_edges, drug_edges)
}
}
network <- list()
network[["source"]] <- ppi_source
network[["complete_network"]] <- list()
network[["complete_network"]][["nodes"]] <- all_nodes
network[["complete_network"]][["edges"]] <- all_edges
if (settings[[ppi_source]]$show_drugs == T) {
network[["complete_network"]][["nodes"]] <- nodes_with_drugs
network[["complete_network"]][["edges"]] <- edges_with_drugs
}
lgr::lgr$info(paste0(
ppi_source_verbose,
": retrieving network communities (",
ppi_source_release,")"))
network[["community_network"]] <-
get_network_communities(edges = all_edges,
nodes = all_nodes)
lgr::lgr$info(paste0(
ppi_source_verbose,
": retrieving hub scores (",
ppi_source_release,")"))
network[["hubscores"]] <-
get_network_hubs(edges = all_edges,
nodes = all_nodes,
genedb = genedb)
} else {
network <- list()
network[["source"]] <- ppi_source
network[["complete_network"]] <- list()
network[["complete_network"]][["nodes"]] <- data.frame()
network[["complete_network"]][["edges"]] <- data.frame()
network[["hubscores"]] <- data.frame()
network[['community_network']] <- NULL
}
return(network)
}
sigven/oncoEnrichR documentation built on Aug. 31, 2023, 8:05 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 get_ppi_network get_string_network_nodes_edges get_biogrid_network_nodes_edges get_network_communities get_network_hubs
get_network_hubs <- function(edges = NULL,
nodes = NULL,
genedb = NULL) {
stopifnot(!is.null(edges) & !is.null(nodes))
validate_db_df(nodes, dbtype = "ppi_nodes")
validate_db_df(edges, dbtype = "ppi_edges")
validate_db_df(genedb, dbtype = "genedb")
edges <- edges |>
dplyr::filter(
!is.na(.data$symbol_A) &
!is.na(.data$symbol_B)) |>
dplyr::select(
c("symbol_A", "symbol_B", "from",
"to", "weight"))
d <- igraph::graph_from_data_frame(d = edges, directed = F)
## hub score (Kleinberg"s hub centrality)
hscore <- igraph::hub_score(d)
hub_scores <- data.frame(
symbol = names(sort(hscore$vector,decreasing = T)),
hub_score = round(sort(hscore$vector,decreasing = T), digits = 3),
stringsAsFactors = F) |>
dplyr::left_join(
dplyr::select(
genedb, c("symbol", "name")),
by = c("symbol"), relationship = "many-to-many") |>
dplyr::left_join(
dplyr::select(
nodes, c("symbol", "query_node")),
by = c("symbol"), relationship = "many-to-many") |>
dplyr::filter(.data$query_node == T) |>
dplyr::select(c("symbol", "name", "hub_score")) |>
dplyr::distinct()
#closeness_score <- igraph::closeness(d, mode="all")
#degree.cent <- centr_degree(d, mode = "all")
return(hub_scores)
}
get_network_communities <- function(edges = NULL, nodes = NULL) {
stopifnot(!is.null(edges) & !is.null(nodes))
validate_db_df(nodes, dbtype = "ppi_nodes")
validate_db_df(edges, dbtype = "ppi_edges")
edges <- edges |>
dplyr::filter(
!is.na(.data$symbol_A) &
!is.na(.data$symbol_B)) |>
dplyr::select(
c("symbol_A","symbol_B",
"from","to","weight"))
d <- igraph::graph_from_data_frame(d = edges, directed = F)
## communities, fast greedy modularity optimization algorithm
## for finding community structure,
cties <- igraph::fastgreedy.community(d)
edges_communities <- data.frame()
if (length(cties) > 0) {
n <- 1
while(n <= length(cties)) {
members_community_n <- cties[[n]]
community <- dplyr::filter(
edges, .data$symbol_A %in% members_community_n &
.data$symbol_B %in% members_community_n) |>
dplyr::mutate(community = n)
edges_communities <- edges_communities |>
dplyr::bind_rows(community)
n <- n + 1
}
}
nodes_communities <- data.frame(
"id" = unique(c(unique(edges_communities$from),
unique(edges_communities$to))),
stringsAsFactors = F) |>
dplyr::inner_join(nodes, by = c("id"), relationship = "many-to-many") |>
dplyr::distinct()
community_structure <- list()
community_structure[["edges"]] <- edges_communities
community_structure[["nodes"]] <- nodes_communities
return(community_structure)
}
get_biogrid_network_nodes_edges <-
function(all_query_nodes = NULL,
add_nodes = 50,
minimum_evidence_items = 2,
show_isolated_nodes = FALSE,
genedb = NULL,
biogrid = NULL) {
stopifnot(!is.null(all_query_nodes))
stopifnot(is.data.frame(all_query_nodes))
stopifnot(NROW(all_query_nodes) > 1)
ppi_edges <- data.frame()
## match edges (interactions) in BioGRID entrezgene_B
edges_part1 <- biogrid |>
dplyr::inner_join(
dplyr::select(all_query_nodes, c("entrezgene","symbol")),
by = c("entrezgene_B" = "entrezgene"),
relationship = "many-to-many"
) |>
dplyr::rename(symbol_B = "symbol") |>
dplyr::left_join(
dplyr::select(
genedb, c("entrezgene", "symbol")),
by = c("entrezgene_A" = "entrezgene"),
relationship = "many-to-many"
) |>
dplyr::rename(symbol_A = "symbol") |>
dplyr::arrange(
.data$symbol_A,
.data$symbol_B,
dplyr::desc(.data$pmid)) |>
dplyr::mutate(
evidence = paste(
.data$method,
.data$throughput,
.data$pmid, sep ="|"
)
)
## match edges (interactions) in BioGRID against entrezgene_B
edges_part2 <- biogrid |>
dplyr::inner_join(
dplyr::select(all_query_nodes, c("entrezgene","symbol")),
by = c("entrezgene_A" = "entrezgene"),
relationship = "many-to-many"
) |>
dplyr::rename(symbol_A = "symbol") |>
dplyr::left_join(
dplyr::select(
genedb, c("entrezgene", "symbol")),
by = c("entrezgene_B" = "entrezgene"),
relationship = "many-to-many"
) |>
dplyr::rename(symbol_B = "symbol") |>
dplyr::arrange(.data$symbol_A,
.data$symbol_B,
dplyr::desc(.data$pmid)) |>
dplyr::mutate(
evidence = paste(
.data$method,
.data$throughput,
.data$pmid, sep ="|"
)
)
all_edges <- as.data.frame(
dplyr::bind_rows(
edges_part1,
edges_part2) |>
dplyr::distinct() |>
dplyr::filter(.data$symbol_A != .data$symbol_B) |>
dplyr::arrange(.data$symbol_A, .data$symbol_B) |>
dplyr::group_by(
.data$entrezgene_A,
.data$entrezgene_B,
.data$symbol_A,
.data$symbol_B) |>
dplyr::summarise(
evidence = paste(
unique(.data$evidence), collapse = ","
),
num_evidence_items = dplyr::n(),
.groups = "drop") |>
## only keep interactions with a minimum number
## of evidence items
dplyr::filter(
.data$num_evidence_items >= minimum_evidence_items) |>
dplyr::mutate(
querynode_A = dplyr::if_else(
.data$entrezgene_A %in% all_query_nodes$entrezgene,
TRUE,
FALSE
),
querynode_B = dplyr::if_else(
.data$entrezgene_B %in% all_query_nodes$entrezgene,
TRUE,
FALSE
)
)
)
## network - query nodes only
ppi_edges <- all_edges |>
dplyr::filter(
.data$querynode_A == T &
.data$querynode_B == T)
if (add_nodes > 0) {
## attach non-query nodes with most interactions with query nodes
edges_single_node_inqueryset <- all_edges |>
dplyr::filter(.data$querynode_A == F |
.data$querynode_B == F)
if (NROW(edges_single_node_inqueryset) > 0) {
set1 <- edges_single_node_inqueryset |>
dplyr::filter(.data$querynode_A == F)
if (NROW(set1) > 0) {
set1 <- set1 |>
dplyr::select(c("entrezgene_A", "entrezgene_B")) |>
dplyr::distinct() |>
dplyr::group_by(.data$entrezgene_A) |>
dplyr::summarise(num_interactions = dplyr::n()) |>
dplyr::arrange(dplyr::desc(.data$num_interactions)) |>
dplyr::rename(entrezgene = "entrezgene_A")
}
set2 <- edges_single_node_inqueryset |>
dplyr::filter(.data$querynode_B == F)
if (NROW(set2) > 0) {
set2 <- set2 |>
dplyr::select(c("entrezgene_A", "entrezgene_B")) |>
dplyr::distinct() |>
dplyr::group_by(.data$entrezgene_B) |>
dplyr::summarise(num_interactions = dplyr::n()) |>
dplyr::arrange(dplyr::desc(.data$num_interactions)) |>
dplyr::rename(entrezgene = "entrezgene_B")
}
if (NROW(set1) > 0 | NROW(set2) > 0) {
all_non_querynodes_ranked <-
dplyr::bind_rows(
set1, set2) |>
dplyr::arrange(
dplyr::desc(.data$num_interactions)) |>
utils::head(add_nodes)
edges_added <-
edges_single_node_inqueryset |>
dplyr::filter(
.data$entrezgene_A %in% all_non_querynodes_ranked$entrezgene |
.data$entrezgene_B %in% all_non_querynodes_ranked$entrezgene
)
if (NROW(edges_added) > 0) {
ppi_edges <- dplyr::bind_rows(
ppi_edges, edges_added
)
}
}
}
}
network <- NULL
## ignore proto-oncogenes/tumor-suppressors with weak
## support/confidence in the protein-protein interaction network
genedb <- genedb |>
dplyr::mutate(oncogene = dplyr::if_else(
.data$oncogene_confidence_level == "MODERATE",
FALSE,
as.logical(.data$oncogene)
)) |>
dplyr::mutate(tumor_suppressor = dplyr::if_else(
.data$tsg_confidence_level == "MODERATE",
FALSE,
as.logical(.data$tumor_suppressor)
))
if(NROW(ppi_edges) > 0){
edges <- ppi_edges |>
dplyr::mutate(from = paste0("s", .data$entrezgene_A)) |>
dplyr::left_join(
dplyr::select(
genedb,
c("entrezgene","oncogene",
"tumor_suppressor",
"cancer_driver")),
by = c("entrezgene_A" = "entrezgene"), relationship = "many-to-many") |>
dplyr::rename(oncogene_A = "oncogene",
tsgene_A = "tumor_suppressor",
cdriver_A = "cancer_driver") |>
dplyr::mutate(to = paste0("s", .data$entrezgene_B)) |>
dplyr::left_join(
dplyr::select(genedb,
c("entrezgene", "oncogene",
"tumor_suppressor","cancer_driver")),
by = c("entrezgene_B" = "entrezgene"), relationship = "many-to-many") |>
dplyr::rename(oncogene_B = "oncogene",
tsgene_B = "tumor_suppressor",
cdriver_B = "cancer_driver") |>
dplyr::mutate(interaction_symbol =
paste0(.data$symbol_A, "_",
.data$symbol_B)) |>
dplyr::mutate(weight = dplyr::if_else(
.data$num_evidence_items >= 5,
as.numeric(2),
as.numeric(1)
)) |>
#dplyr::mutate(weight = .data$num_evidence_items) |>
dplyr::mutate(title = paste0("Number of evidence items:",
.data$num_evidence_items)) |>
dplyr::distinct()
nodes <- data.frame(
"entrezgene" = unique(c(edges$entrezgene_A,
edges$entrezgene_B)),
stringsAsFactors = F) |>
dplyr::distinct() |>
dplyr::left_join(dplyr::select(
genedb,
c("symbol","entrezgene",
"tumor_suppressor","oncogene",
"cancer_driver","genename",
"targeted_cancer_drugs_ep",
"targeted_cancer_drugs_lp")),
by = "entrezgene", relationship = "many-to-many") |>
dplyr::filter(!is.na(.data$entrezgene)) |>
dplyr::left_join(
dplyr::select(all_query_nodes,
c("entrezgene","query_node")),
by = "entrezgene", relationship = "many-to-many") |>
dplyr::mutate(id = paste0("s", .data$entrezgene)) |>
dplyr::distinct() |>
dplyr::mutate(query_node = dplyr::if_else(
is.na(.data$query_node),
FALSE,
as.logical(.data$query_node)))
if(show_isolated_nodes == T){
nodes <- nodes |>
dplyr::bind_rows(all_query_nodes) |>
dplyr::distinct() |>
dplyr::mutate(query_node = dplyr::if_else(
is.na(.data$query_node),
FALSE,
as.logical(.data$query_node)))
}
network <- list()
network$edges <- edges
network$nodes <- nodes
}
return(network)
}
get_string_network_nodes_edges <-
function(qgenes,
all_query_nodes = NULL,
query_type = "network",
minimum_score = 0.9,
add_nodes = 50,
show_isolated_nodes = FALSE,
network_type = "physical",
genedb = NULL) {
query_list <- paste(qgenes, collapse = "%0d")
edges <- jsonlite::fromJSON(
paste0("https://string-db.org/api/json/",
query_type,
"?species=9606&identifiers=", query_list,
"&required_score=", as.integer(minimum_score * 1000),
"&network_type=", network_type,
"&add_nodes=", add_nodes))
## ignore proto-oncogenes/tumor-suppressors with weak
## support/confidence in the protein-protein interaction network
genedb <- genedb |>
dplyr::mutate(oncogene = dplyr::if_else(
.data$oncogene_confidence_level == "MODERATE",
FALSE,
as.logical(.data$oncogene)
)) |>
dplyr::mutate(tumor_suppressor = dplyr::if_else(
.data$tsg_confidence_level == "MODERATE",
FALSE,
as.logical(.data$tumor_suppressor)
))
if (NROW(edges) > 0) {
edges <- edges |>
dplyr::rename(symbol_A = "preferredName_A",
symbol_B = "preferredName_B") |>
dplyr::left_join(
dplyr::select(genedb, c("entrezgene", "symbol")),
by = c("symbol_A" = "symbol"), relationship = "many-to-many") |>
dplyr::filter(!is.na(.data$entrezgene)) |>
dplyr::rename(entrezgene_A = "entrezgene") |>
dplyr::mutate(from = paste0("s", .data$entrezgene_A)) |>
dplyr::left_join(
dplyr::select(
genedb,
c("entrezgene","oncogene",
"tumor_suppressor",
"cancer_driver")),
by = c("entrezgene_A" = "entrezgene"), relationship = "many-to-many") |>
dplyr::rename(oncogene_A = "oncogene",
tsgene_A = "tumor_suppressor",
cdriver_A = "cancer_driver") |>
dplyr::left_join(dplyr::select(genedb, c("entrezgene", "symbol")),
by = c("symbol_B" = "symbol"), relationship = "many-to-many") |>
dplyr::filter(!is.na(.data$entrezgene)) |>
dplyr::rename(entrezgene_B = "entrezgene") |>
dplyr::mutate(to = paste0("s", .data$entrezgene_B)) |>
dplyr::left_join(
dplyr::select(genedb,
c("entrezgene", "oncogene",
"tumor_suppressor","cancer_driver")),
by = c("entrezgene_B" = "entrezgene"), relationship = "many-to-many") |>
dplyr::rename(oncogene_B = "oncogene",
tsgene_B = "tumor_suppressor",
cdriver_B = "cancer_driver") |>
dplyr::mutate(interaction_symbol =
paste0(.data$symbol_A, "_",
.data$symbol_B)) |>
dplyr::left_join(
dplyr::select(all_query_nodes, c("symbol","query_node")),
by = c("symbol_A" = "symbol"), relationship = "many-to-many") |>
dplyr::rename(querynode_A = "query_node") |>
dplyr::left_join(
dplyr::select(all_query_nodes, c("symbol","query_node")),
by = c("symbol_B" = "symbol"), relationship = "many-to-many") |>
dplyr::rename(querynode_B = "query_node") |>
dplyr::mutate(
querynode_A = dplyr::if_else(
is.na(.data$querynode_A),
FALSE,
as.logical(.data$querynode_A)
),
querynode_B = dplyr::if_else(
is.na(.data$querynode_B),
FALSE,
as.logical(.data$querynode_B)
)
) |>
dplyr::mutate(weight = .data$score) |>
dplyr::mutate(title = paste0("Interaction score:", .data$score)) |>
dplyr::distinct() |>
dplyr::select(-c("ncbiTaxonId", "stringId_A", "stringId_B"))
nodes <- data.frame(
"symbol" = unique(c(edges$symbol_A, edges$symbol_B)),
stringsAsFactors = F) |>
dplyr::distinct() |>
dplyr::left_join(dplyr::select(
genedb,
c("symbol","entrezgene",
"tumor_suppressor","oncogene",
"cancer_driver","genename",
"targeted_cancer_drugs_ep",
"targeted_cancer_drugs_lp")),
by = "symbol", relationship = "many-to-many") |>
dplyr::filter(!is.na(.data$entrezgene)) |>
dplyr::left_join(
dplyr::select(all_query_nodes,
c("symbol","query_node")),
by = "symbol", relationship = "many-to-many") |>
dplyr::mutate(id = paste0("s", .data$entrezgene)) |>
dplyr::distinct() |>
dplyr::mutate(query_node = dplyr::if_else(
is.na(.data$query_node),
FALSE,
as.logical(.data$query_node)))
if(show_isolated_nodes == T){
nodes <- nodes |>
dplyr::bind_rows(all_query_nodes) |>
dplyr::distinct() |>
dplyr::mutate(query_node = dplyr::if_else(
is.na(.data$query_node),
FALSE,
as.logical(.data$query_node)))
}
network <- list()
network$edges <- edges
network$nodes <- nodes
}
else{
network <- NULL
}
return(network)
}
get_ppi_network <- function(qgenes = NULL,
ppi_source = "string",
genedb = NULL,
cancerdrugdb = NULL,
biogrid = NULL,
settings = NULL,
ppi_source_release = NA) {
lgr::lgr$appenders$console$set_layout(
lgr::LayoutFormat$new(timestamp_fmt = "%Y-%m-%d %T"))
stopifnot(!is.null(qgenes))
stopifnot(!is.null(settings))
stopifnot(!is.null(genedb))
validate_db_df(genedb, dbtype = "genedb")
stopifnot(is.integer(qgenes))
stopifnot(!is.null(cancerdrugdb))
stopifnot(!is.null(biogrid))
validate_db_df(biogrid, dbtype = "biogrid")
stopifnot(!is.null(cancerdrugdb[['network']]))
stopifnot(ppi_source == "string" | ppi_source == "biogrid")
if (ppi_source == "string") {
stopifnot(identical(
names(settings[[ppi_source]]),
c("minimum_score",
"visnetwork_shape",
"visnetwork_shadow",
"show_drugs",
"add_nodes",
"query_type",
"network_type",
"show_isolated_nodes"))
)
stopifnot(settings[[ppi_source]]$query_type == "interaction_partners" |
settings[[ppi_source]]$query_type == "network")
stopifnot(settings[[ppi_source]]$network_type == "physical" |
settings[[ppi_source]]$network_type == "functional")
}
if (ppi_source == "biogrid") {
stopifnot(identical(
names(settings[[ppi_source]]),
c("minimum_evidence",
"visnetwork_shape",
"visnetwork_shadow",
"show_drugs",
"add_nodes",
"show_isolated_nodes"))
)
}
query_nodes <- data.frame("entrezgene" = qgenes, stringsAsFactors = F) |>
dplyr::distinct() |>
dplyr::left_join(dplyr::select(
genedb,
c("symbol","entrezgene","tumor_suppressor",
"cancer_driver", "oncogene", "genename",
"targeted_cancer_drugs_ep",
"targeted_cancer_drugs_lp")),
by = c("entrezgene" = "entrezgene"), relationship = "many-to-many") |>
dplyr::mutate(id = paste0("s", .data$entrezgene)) |>
dplyr::mutate(query_node = T) |>
dplyr::distinct()
## Occasionally, some entrezgene identifiers are ambiguous (wrt. gene symbols)
## if this is the case, pick random entry from duplicate records, so that
## all entrezgene identifiers point to a unique row
if (nrow(query_nodes) > length(qgenes)) {
duplicate_records <- as.data.frame(
query_nodes |>
dplyr::group_by(.data$entrezgene) |>
dplyr::summarise(n = dplyr::n(),
.groups = "drop") |>
dplyr::filter(.data$n > 1)
)
lgr::lgr$info( paste0(
"Resolving ambiguous Entrez gene identifiers: ",
paste(duplicate_records$entrezgene, collapse = ", ")))
query_nodes_clean <-
as.data.frame(
query_nodes |>
dplyr::group_by(.data$entrezgene) |>
dplyr::summarise(n = dplyr::n(),
.groups = "drop") |>
dplyr::filter(.data$n == 1)
)
for (i in 1:nrow(duplicate_records)) {
entrezgene_id = duplicate_records[i,]$entrezgene
sample_row <- dplyr::sample_n(
dplyr::filter(query_nodes, .data$entrezgene == entrezgene_id),
1
)
query_nodes_clean <- query_nodes_clean |>
dplyr::bind_rows(sample_row)
i <- i + 1
}
if (length(qgenes) == nrow(query_nodes_clean)) {
query_nodes <- query_nodes_clean
}
}
ppi_source_verbose <- "BioGRID"
if (ppi_source == "string") {
ppi_source_verbose <- "STRING"
lgr::lgr$info( paste0(
ppi_source_verbose,
": retrieving protein-protein interaction network from API (",
ppi_source_release,")"))
lgr::lgr$info( paste0(
ppi_source_verbose,
": Settings - required minimum score = ",
settings[[ppi_source]]$minimum_score,
", add_nodes = ",settings[[ppi_source]]$add_nodes,
", show_isolated_nodes = ",settings[[ppi_source]]$show_isolated_nodes,
", network_type = ", settings[[ppi_source]]$network_type))
all_edges <- data.frame()
all_nodes <- data.frame()
if (length(qgenes) > 200) {
i <- 1
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,
query_type = settings[[ppi_source]]$query_type,
network_type = settings[[ppi_source]]$network_type,
minimum_score = settings[[ppi_source]]$minimum_score,
add_nodes = settings[[ppi_source]]$add_nodes,
show_isolated_nodes = settings[[ppi_source]]$show_isolated_nodes,
genedb = genedb)
all_edges <- all_edges |>
dplyr::bind_rows(ppi_network_data$edges) |>
dplyr::distinct()
all_nodes <- all_nodes |>
dplyr::bind_rows(ppi_network_data$nodes) |>
dplyr::distinct()
i <- i + 199
}
} else {
ppi_network_data <- get_string_network_nodes_edges(
qgenes = qgenes,
all_query_nodes = query_nodes,
query_type = settings[[ppi_source]]$query_type,
network_type = settings[[ppi_source]]$network_type,
minimum_score = settings[[ppi_source]]$minimum_score,
show_isolated_nodes = settings[[ppi_source]]$show_isolated_nodes,
add_nodes = settings[[ppi_source]]$add_nodes,
genedb = genedb)
if (!is.null(ppi_network_data)) {
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()
}
}
}
if (ppi_source == "biogrid") {
lgr::lgr$info(paste0(
ppi_source_verbose,
": retrieving physical protein-protein interaction network (",
ppi_source_release,")"))
lgr::lgr$info( paste0(
ppi_source_verbose,
": Settings - minimum evidence = ",
settings[[ppi_source]]$minimum_evidence,
", show_isolated_nodes = ",settings[[ppi_source]]$show_isolated_nodes,
", add_nodes = ",settings[[ppi_source]]$add_nodes))
all_edges <- data.frame()
all_nodes <- data.frame()
ppi_network_data <- get_biogrid_network_nodes_edges(
all_query_nodes = query_nodes,
minimum_evidence_items = settings[[ppi_source]]$minimum_evidence,
add_nodes = settings[[ppi_source]]$add_nodes,
show_isolated_nodes = settings[[ppi_source]]$show_isolated_nodes,
biogrid = biogrid,
genedb = genedb)
if (!is.null(ppi_network_data)) {
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()
}
}
if (nrow(all_edges) > 0 & nrow(all_nodes) > 0) {
all_nodes <- all_nodes |>
dplyr::mutate(shape = dplyr::if_else(
.data$query_node == T,
settings[[ppi_source]]$visnetwork_shape,
as.character("box")))
all_nodes$shadow <- settings[[ppi_source]]$visnetwork_shadow
all_nodes$title <- unlist(
lapply(stringr::str_match_all(
all_nodes$genename,">.+<"), paste, collapse = ","))
all_nodes$title <- stringr::str_replace_all(
all_nodes$title,">|<", "")
all_nodes$label <- all_nodes$symbol # Node label
all_edges$width <- all_edges$weight
all_nodes$gene_category <- "protein_coding"
all_nodes$size <- 25
all_nodes <- all_nodes |>
dplyr::mutate(color.background = dplyr::if_else(
.data$query_node == T, "lightblue", "mistyrose")) |>
dplyr::mutate(color.background = dplyr::if_else(
.data$tumor_suppressor == T & .data$oncogene == F,
"firebrick", as.character(.data$color.background),
as.character(.data$color.background))) |>
dplyr::mutate(color.background = dplyr::if_else(
.data$oncogene == T & .data$tumor_suppressor == F,
"darkolivegreen", as.character(.data$color.background),
as.character(.data$color.background))) |>
dplyr::mutate(color.background = dplyr::if_else(
.data$oncogene == T & .data$tumor_suppressor == T,
"black", as.character(.data$color.background),
as.character(.data$color.background))) |>
dplyr::mutate(color.border = "black",
color.highlight.background = "orange",
color.highlight.border = "darkred",
font.color = "white") |>
dplyr::mutate(font.color = dplyr::if_else(
.data$query_node == T | .data$color.background == "mistyrose",
"black", as.character(.data$font.color),
as.character(.data$font.color)))
nodes_with_drugs <- all_nodes
edges_with_drugs <- all_edges
if (settings[[ppi_source]]$show_drugs == T) {
drug_target_ids <-
dplyr::inner_join(
dplyr::select(all_nodes, c("id")),
dplyr::select(cancerdrugdb[['network']]$edges, c("from")),
by = c("id" = "from"), relationship = "many-to-many") |>
dplyr::distinct()
if (nrow(drug_target_ids) > 0) {
drug_edges <- dplyr::inner_join(
cancerdrugdb[['network']]$edges,
drug_target_ids, by = c("from" = "id"))
drug_nodes <- drug_edges |>
dplyr::select(-c("from")) |>
dplyr::inner_join(cancerdrugdb[['network']]$nodes,
by = c("to" = "id"), relationship = "many-to-many") |>
dplyr::rename(id = "to") |>
dplyr::distinct()
nodes_with_drugs <- dplyr::bind_rows(all_nodes, drug_nodes)
edges_with_drugs <- dplyr::bind_rows(all_edges, drug_edges)
}
}
network <- list()
network[["source"]] <- ppi_source
network[["complete_network"]] <- list()
network[["complete_network"]][["nodes"]] <- all_nodes
network[["complete_network"]][["edges"]] <- all_edges
if (settings[[ppi_source]]$show_drugs == T) {
network[["complete_network"]][["nodes"]] <- nodes_with_drugs
network[["complete_network"]][["edges"]] <- edges_with_drugs
}
lgr::lgr$info(paste0(
ppi_source_verbose,
": retrieving network communities (",
ppi_source_release,")"))
network[["community_network"]] <-
get_network_communities(edges = all_edges,
nodes = all_nodes)
lgr::lgr$info(paste0(
ppi_source_verbose,
": retrieving hub scores (",
ppi_source_release,")"))
network[["hubscores"]] <-
get_network_hubs(edges = all_edges,
nodes = all_nodes,
genedb = genedb)
} else {
network <- list()
network[["source"]] <- ppi_source
network[["complete_network"]] <- list()
network[["complete_network"]][["nodes"]] <- data.frame()
network[["complete_network"]][["edges"]] <- data.frame()
network[["hubscores"]] <- data.frame()
network[['community_network']] <- NULL
}
return(network)
}
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.