Function to download biotic interaction datasets
[This article was first published on R Code – Geekcologist, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
I work in ecology, biogeography, etc… Biotic interactions (interactions between species) and its repercussions on species distributions is my main research interest.
As such, I had, at some point, to download datasets on species interactions. I wanted to be able to produce a uniform (more or less, not as much as I would like) R object whatever the database I used. I created this function to do just that!
It can be used to download from database website or upload csv files into R (depending on the database). It currently uses the following databases:
- EcoBase – The database for the models using Ecopath with Ecosim. These are really good, have good metadata and spatial information (polygons) (Christensen & Walters, 2004; Heymans et al. 2016).
- Web of Life – To be really honest I don’t know this database all that much. However it also provides information on the dataset location, but it does not provide details on the ecosystem where it occurs (Fortuna et al. 2014).
- Global Web – This database provides details on the ecosystem where the interaction dataset occur, bu no details on spatial location are provided (Thompson et al. 2012).
- Mangal – This database provides lots of information on the biotic interactions datasets (Poisot et al. 2015).
In order to write this function I relied on work previously done by others. Namely the code to download EcoBase interaction datasets was obtained here. In what concerns the mangal database I relied on the work done in the package rmangal.
NOTE: This is important! Be sure to check the results of the function! I’m still working on it, so this is a provisional version. But I think it works pretty good though.
(Somewhere in the future it might be available in my GitHub.)
create.fw.list <- function(db, folder = NULL, type = NULL, ecosyst=FALSE, ref=FALSE, spatial=FALSE, code=FALSE)
{
#### Arguments ####
#'db' - database - eb (EcoBase), gw (GlobalWeb), wl (Web of Life) and mg (Mangal)
#'folder' - folder in the WD to get the dataset files (db=gw and wl).
#'type' - if db=mg the user should provide the type of interactions to be downloaded
# 'ecosyst' - Getting ecosystem information, only for gw, eb
#'ref' references information
#'spatial' - get spatial info, only for wl, eb and mg
#### Data Sources ####
#Global Web: https://www.globalwebdb.com/
#EcoTroph Example (EcoBase): http://sirs.agrocampus-ouest.fr/EcoTroph/index.php?action=examples
#Script for EcoBase: http://sirs.agrocampus-ouest.fr/EcoBase/#discoverytools
#Mangal: #https://mangal-wg.github.io/rmangal/articles/rmangal.html
#Web of Life: http://www.web-of-life.es/
#### Results ####
fwlist <- list()
#### Conditions for data entry ############################################################
#db
if(!db %in% c("eb","wl","gw","mg"))stop("Argument 'db' must take one of the following values:\n
'wl' - Web of Life
'mg' - mangal
'gw' - globalweb
'eb' - ecobase")
#folder
if(!db %in% c("wl","gw") & !is.null(folder)) stop("Argument 'folder'can only be used if 'db'= 'wl' or 'gw'!")
#type
#folder
if(!db %in% c("mg") & !is.null(type)) stop("Argument 'type'can only be used if 'db'= 'mg'!")
#ecosyst
if(!db %in% c("gw", "eb") & ecosyst==TRUE) stop("Argument 'ecosyst'can only be used if 'db'= 'eb' or 'gw'!")
#ref
#all
#spatial
if(!db %in% c("wl", "mg", "eb") & spatial==TRUE) stop("Argument 'spatial'can only be used if 'db'= 'eb', 'mg' 'wl'!")
#code
if(!db %in% c("wl", "mg", "gw") & code==TRUE) stop("Argument 'code'can only be used if 'db'= 'wl', 'mg'', 'gw'!")
############################################################################################
#Updating each dataset database
#GlobalWeb
if (db == "gw"){
message("####################### GLOBALWEB DATABASE #######################\n\n")
message("Fetching info from the provided folder!")
files_gw <- list.files(path = folder, pattern = "WEB")
ngw <- length(files_gw)
message (paste0("There are ", ngw, " food web files in the folder!"))
#Load files into list
#And create vector of references
if (ref==TRUE) reflist_gw <- c()
names_gw <- c()#FW names
#getting the files into R
for(i in 1:ngw){
message(paste0("Fetching food web ", i, " in ", ngw, "!"))
dfgw <- read.csv(paste0(folder,"/",files_gw[i]), header = FALSE) # read csv file
dfgw <- dfgw[, colSums( is.na(dfgw) ) <=1]#Remove columns with all NA
#Get the FW name
names_gw[i] <- as.character(dfgw[2,1])
#Get the reference to the vector
if (ref==TRUE) reflist_gw[i] <- as.character(dfgw[1,1])
#to name the columns
names_gw_c <- c()
n1 <- ncol(dfgw)-1
for(j in 1:n1){
names_gw_c[j] <- as.character(dfgw[2,j+1])
}
#to name the rows
names_gw_r <- c()
n2 <- nrow(dfgw)-2
for(j in 1:n2){
names_gw_r[j] <- as.character(dfgw[j+2, 1])
}
dfgw <- dfgw[-c(1,2),-1]
#Remove columns with NA
dfgw[dfgw==""] <- NA
dfgw <- na.omit(dfgw)
if(i==281){names_gw_r <- names_gw_r[-c(36,37)]}#the FW on i=281 has a note at the bottom
#Delete the 'empty names'
names_gw_c <- names_gw_c[names_gw_c!=""]
names_gw_r <- names_gw_r[names_gw_r!=""]
#Same names in rows or columns?
#if(length(unique(names_gw_r)) < length(names_gw_r)) rown[i] <- as.character(i)
#if(length(unique(names_gw_c)) < length(names_gw_c)) coln[i] <- as.character(i)
#For some strange reason some rows and columns have the same name
names_gw_c <- paste0("sp_", as.character(1:length(names_gw_c)), "_",names_gw_c)
names_gw_r <- paste0("sp_", as.character(1:length(names_gw_r)), "_",names_gw_r)
colnames(dfgw) <- names_gw_c
rownames(dfgw) <- names_gw_r
fwlist[[i]] <- dfgw
}
#Name the list
names(fwlist) <- names_gw
if(ref==TRUE){
references <- as.data.frame(matrix(ncol = 4))
names(references) <- c("FW code", "first_author", "year", "full_ref" )
files_gw <- list.files(folder, pattern = "WEB")
message("Fetching references from the dataset files!")
for(w in 1:ngw){
dfgw <- read.csv(paste0(folder,"/",files_gw[w]), header = FALSE) # read csv file
#message(paste0("Reading file ", files_gw[w]))
dfgw <- dfgw[, colSums( is.na(dfgw) ) <=1]#Remove columns with all NA
#Get the reference to the vector
full_ref1 <- as.character(dfgw[1,1])
references[w,4] <- full_ref1#full reference
references[w,1] <- files_gw[w]#fw code
references[w,2] <- str_sub(word(full_ref1, start = 1), 1, str_length(word(full_ref1, start = 1))-1)#fisrt author
references[w,3] <- regmatches(x = full_ref1,gregexpr("[0-9]+",text = full_ref1))[[1]][1]#year
#references[w,3] <- gsub('.+\\(([0-9]+)\\).+?$', '\\1', full_ref1)#year
}#end loop to add refs
}#end gw refs
#ECOSYSTEM
if(ecosyst==TRUE){
message("Searching for 'gw_list.csv' file...")
if (!file.exists(paste0(folder, "/gw_list.csv"))) stop("\nThe pdf 'gw_list.pdf' has to be previously converted to a csv file...")
#I had to conver the gw_list.pdf file to excel (csv), since I could not install tabulizes to extract pdf tables
gw_eco <- read.csv(paste0(folder,"/","gw_list.csv"), header = TRUE, sep = ";") # read csv file
filn <- paste0("WEB", as.character(gw_eco[,1]), ".csv")
gw_eco2 <- gw_eco[,1:3]
gw_eco2[,1] <- filn
names(gw_eco2)[1] <- "FW"
#yes... I do know the following few lines are 'ugly'...
filn <- as.data.frame(cbind(filn, filn))
names(filn) <- c("filn1","filn2")
#files_gw <- list.files(path = folder, pattern = "WEB")
ecosystem <- merge(x=filn, y=gw_eco2, by.x= "filn2", by.y = "FW")
ecosystem <- ecosystem[,c(2, 3, 4)]
names(ecosystem)[1] <- "Food web"
}
}#end of gw
#Web of Life
if (db == "wl"){
message("####################### WEB OF LIFE DATABASE #######################\n\n")
files_wl <- list.files(path = folder, pattern = "FW")
nwl <- length(files_wl)
message (paste0("There are ", nwl, " food web files in the folder!"))
#Get refs and metrics table
if (file.exists(paste0(folder, "/references.csv"))) {
table_wl <- read.csv(paste0(folder, "/references.csv"), header = TRUE) # read csv file
} else {
stop("There is no 'references.csv' file on the folder, as provided by the website!")
}
#FW names
names_wl <- as.character(table_wl[,8])
#Load files
for(i in 1:nwl){
message(paste0("Fetching food web ", i, " in ", nwl, "!"))
dfwl <- read.csv(paste0(folder, "/",files_wl[i]), header = TRUE) # read csv file
#row.names(dfwl) <- as.character(dfwl[,1])
#dfwl <- dfwl[,-1]
dfwl[is.na(dfwl)] <- 0
fwlist[[i]] <- dfwl
}
names(fwlist) <- names_wl
#REFERENCES
if(ref==TRUE){
references <- as.data.frame(matrix(ncol = 4))
names(references) <- c("FW code", "first_author", "year", "full_ref" )
message("Fetching references from the 'references.csv' file!")
message("Checking the presence of the 'references.csv' file...")
if(!file.exists(paste0(folder, "/references.csv"))==TRUE)stop("Can't retrieve reference details... \n File not present!")
ref_file <- read.csv(paste0(folder, "/references.csv"), header = TRUE) # read csv file
for(w in 1:nwl){
full_ref1 <- as.character(ref_file[w,7])
references[w,4] <- full_ref1#full reference
references[w,1] <- as.character(ref_file[w,1])#fw code
references[w,2] <- str_sub(word(full_ref1, start = 1), 1, str_length(word(full_ref1, start = 1))-1)#fisrt author
references[w,3] <- regmatches(x = full_ref1,gregexpr("[0-9]+",text = full_ref1))[[1]][1]#year
#references[w,3] <- gsub('.+\\(([0-9]+)\\).+?$', '\\1', full_ref1)#year
}#end loop to add refs
}#end wl refs
#SPATIAL
if(spatial==TRUE){
message("Fetching the spatial information from the 'references.csv' file!")
message("Checking the presence of the 'references.csv' file...")
if(!file.exists(paste0(folder, "/references.csv"))==TRUE)stop("Can't retrieve spatial info... \n File not present!")
ref_file <- read.csv(paste0(folder, "/references.csv"), header = TRUE) # read csv file
spatial1 <- ref_file[,c(1,9,10)]
}#end of spatial
}#end of wl
#EcoBase
if(db == "eb"){
message("####################### ECOBASE DATABASE #######################\n\n")
message("Fetching info from the EcoBase website!")
suppressWarnings({
#To obtain the list of available models
suppressMessages({
h=basicTextGatherer()
curlPerform(url = 'http://sirs.agrocampus-ouest.fr/EcoBase/php/webser/soap-client_3.php',writefunction=h$update)
data1 <- xmlTreeParse(h$value(),useInternalNodes=TRUE)
liste_mod <- ldply(xmlToList(data1),data.frame)#liste_mod contains a list and decription
})
#Select only those allowing dissemination
l2 <- subset(liste_mod, model.dissemination_allow =="true")#only those of which dissemination is allowed
message("Sellected only those to which model dissemination is allowed!")
#Select only those with whole food webs
l3 <- subset(l2, model.whole_food_web =="true")#only those with the full food web
message("Sellected only those to which the whole food web is available!")
#Get model names
model.name <- as.character(l3$model.model_name)
input_list <- list()
id <- as.numeric(as.character(l3$model.model_number))
#Loop to get input list
for(i in 1:nrow(l3)){
message(paste0("Fetching information on food web ",i, " of ", nrow(l3)))
suppressMessages({
h=basicTextGatherer()
mymodel <- id[i]
curlPerform(url = paste('http://sirs.agrocampus-ouest.fr/EcoBase/php/webser/soap-client.php?no_model=',mymodel,sep=''),writefunction=h$update,verbose=TRUE)
data2 <- xmlTreeParse(h$value(),useInternalNodes=TRUE)
input1 <- xpathSApply(data2,'//group',function(x) xmlToList(x))
})
#need do name the columns
names_input <- as.character(input1[1,])
input1 <- as.data.frame(input1)
colnames(input1) <- names_input
input1 <- input1[-1,]
input_list[[i]] <- input1
}#end of loop to get input list
mnames <- names(input_list)
for (i in 1:length(input_list)){
m2 <- input_list[[i]] #get the model
nnodes <- length(m2)
node_names <- names(m2)
# if (biomass == TRUE)
# {
# nodes_biomass <- as.data.frame(matrix(ncol=3, nrow=nnodes))
# names(nodes_biomass) <- c("id", "name", "biomass")
# }
int_matrix <- as.data.frame(matrix(ncol=nnodes, nrow=nnodes))
for(j in 1:length(m2)){
node1 <- m2[[j]]
node_id <- as.numeric(node1$group_seq)
#node1_biomass <- as.numeric(node1$biomass)
node_name <- node_names[j]
#biomass
#if (biomass == TRUE)
#{
#nodes_biomass[node_id, 1] <- node_id
#nodes_biomass[node_id, 2] <- node_name
#nodes_biomass[node_id, 3] <- node1_biomass
#}
#matrix
colnames(int_matrix)[node_id] <- node_name
rownames(int_matrix)[node_id] <- node_name
diet_node1 <- node1$diet_descr
nr_food_items <- length(diet_node1)
for(a in 1:nr_food_items){
item1 <- diet_node1[[a]]
id_item1 <- as.numeric(item1$prey_seq)
proportion_item1 <- as.numeric(item1$proportion)
detritus_item1 <- as.numeric(item1$detritus_fate)
#send to matrix
int_matrix[id_item1,node_id] <- proportion_item1
}
}
int_matrix[is.na(int_matrix)] <- 0#replacing NA with 0
#if(db=="eb" && biomass == TRUE) fwlist[[i]] <- list(biomass=nodes_biomass, trophic_relations=int_matrix)
#if(db=="eb" && biomass == FALSE) fwlist[[i]] <- int_matrix
fwlist[[i]] <- int_matrix
}
names(fwlist) <- model.name
})#end of outer suppressWarnings
#REFERENCES
if(ref==TRUE){
references <- as.data.frame(matrix(ncol = 4))
names(references) <- c("FW code", "first_author", "year", "full_ref" )
message("Fetching the references information!")
for(w in 1:nrow(l3)){
#Get the reference to the vector
full_ref1 <- as.character(l3$model.reference)[w]
references[w,4] <- full_ref1#full reference
references[w,1] <- as.numeric(as.character(l3$model.model_number[w]))#fw code
references[w,2] <- as.character(l3$model.author[w])#fisrt author
references[w,3] <- regmatches(x = full_ref1,gregexpr("[0-9]+",text = full_ref1))[[1]][1]#year
#references[w,3] <- gsub('.+\\(([0-9]+)\\).+?$', '\\1', full_ref1)#year
}#end loop to add refs
}#end of eb refs
#ECOSYSTEM
if(ecosyst==TRUE){
ecosystem <- data.frame(l3$model.model_number, l3$model.country, l3$model.ecosystem_type)
names(ecosystem) <- c("Food web", "Location", "Ecosystem")
}#end of eb ecosystem
#SPATIAL
if(spatial==TRUE){
message("Fetching spatial information from the EcoBase website...")
#Get actual polygons
EcoBase_shape <- sf::st_read("http://sirs.agrocampus-ouest.fr/EcoBase/php/protect/extract_kml.php")
ebd <- EcoBase_shape$Name
#Getting the model numbers
nmr <- list()
for(i in 1:length(ebd)){
nr <- strsplit(as.character(ebd[i]), "--::")[[1]][1]
nr <- as.numeric(str_extract_all(nr, "\\d+")[[1]])#Alternative to Numextract
nmr[[i]] <- nr
}
nmr2 <- c()#line rows for each model
for(i in 1:length(nmr)){
a <- nmr[[i]]
b <- length(a)
c1 <- rep(i,b)
nmr2 <- c(nmr2, c1)
}
#In Which row in ecobase geo file is the model?
nmr <- unlist(nmr)
table1 <- as.data.frame(cbind(nmr2, nmr))
colnames(table1) <- c("row_n","id")
#In which row does model.model_number with a given Id occurs?
lines_n <- c()
for (i in 1:nrow(liste_mod)){
id <- as.numeric(as.character(liste_mod$model.model_number[i]))
lines_n[i] <- as.numeric(table1[table1$id==id,][1])
}
ecobase_poly2 <- list()
for(i in 1:length(lines_n)){
ecobase_poly2[i] <- st_geometry(EcoBase_shape)[lines_n[i]]
#plot(st_geometry(EcoBase_shape)[lines_n[i]], border="green", add=TRUE)
}
#if no polygon then bounding box
#into here ecobase_poly2
for(i in 1:length(ecobase_poly2)){
if(is.na(lines_n[i])){
#create a bounding box geographic thing
z1 <- as.numeric(Numextract(liste_mod$model.geographic_extent[[i]]))
z2 <- c(z1[4], z1[1], z1[2], z1[1], z1[2], z1[3], z1[4], z1[3])
x1 <- as.data.frame(matrix(z2, ncol=2, byrow=TRUE))
x1 <- cbind(x1[2], x1[1])#had to change lat and long... I had this the other way around...
p1 <- Polygon(x1)
ps1 <- Polygons(list(p1),1)
ecobase_poly2[[i]] <- st_as_sf(SpatialPolygons(list(ps1)))
}
ecobase_poly2[[i]] <- ecobase_poly2[[i]]
}
#convert all to class sf
for(i in 1:length(ecobase_poly2)){
if(!any(class(ecobase_poly2[[i]])=='sf')){
t2 <- ecobase_poly2[[i]]
t3 <- st_cast(t2, to="POLYGON")
ecobase_poly2[[i]] <- st_as_sf(as(st_zm(st_geometry(t3)), "Spatial"))
}
else message("Ok!")
}
#line.Id correspondence
table2 <- as.data.frame(cbind(1:length(ecobase_poly2),as.numeric(as.character(liste_mod$model.model_number))))
names(table2) <- c("row","id")
#select the corresponding polygons
id_selected <- as.numeric(as.character(l3$model.model_number))
#Which rows?
rows_selected <- c()
for(i in 1:length(id_selected)){
rows_selected[i] <- as.numeric(table2[table2["id"]==id_selected[i],][1])
}
spatial1 <- ecobase_poly2[rows_selected]
}#end of eb spatial
}#end of eb
#MANGAL
if(db == "mg"){
message("####################### MANGAL DATABASE #######################\n\n")
message("Fetching datasets from the Mangal website! \n\n Types 'predation' and 'herbivory' by default... \n but run mangal function 'avail_type' to check available types...\n\nThis operation might take a long time!")
ntypes <- length(type)
net_info <- list()
for(i in 1:ntypes){
message(paste0("\n\nFetching information from interactions of the type ","'",type[i], "'!"))
fwlist1 <- search_interactions(type = type[i]) %>% get_collection()
net_info <- rbind(net_info, fwlist1)
fwlist2 <- as.igraph(fwlist1)
fwlist <- c(fwlist, fwlist2)
#class(fwlist)
}
#Converting igraph objects to data frame
for(i in 1:length(fwlist)){
fw2 <- fwlist[[i]]
#convert each igraph to a data frame
fw3 <- as_data_frame(fw2, what = "both")
id_name <- fw3$vertices[,1:2]
for(j in 1:nrow(id_name)){#clean the names
node_name <- id_name$original_name[j]
if (grepl(":", node_name, fixed=TRUE)) {
node_name <- tail(strsplit(node_name, ": "))[[1]]
id_name[j,2] <- node_name[2]
} else id_name[j,2] <- node_name
}#end clean names
id_edges <- fw3$edges[,1:3]
int_matrix <- as.data.frame(matrix(ncol = nrow(id_name), nrow = nrow(id_name)))
colnames(int_matrix) <- id_name$original_name
rownames(int_matrix) <- id_name$original_name
#Fill the matrix
for(a in 1:nrow(id_edges)){
edge1 <- as.numeric(id_edges[a,1:2])
name1 <- id_name[as.character(edge1[1]),][,2]
name2 <- id_name[as.character(edge1[2]),][,2]
int_matrix[name1,name2] <- 1
}
int_matrix[is.na(int_matrix)] <- 0 #convert all NA to zero
fwlist[[i]] <- int_matrix
}#end of loop to convert to a data frame
if(ref==TRUE){
references <- as.data.frame(matrix(ncol = 4))
names(references) <- c("Dataset ID", "first_author", "year", "DOI" )
message("Fetching references!")
for(j in 1:length(net_info)){
dataset_id <- net_info[[j]]$dataset$dataset_id
first_author <- net_info[[j]]$reference$first_author
year_mng <- as.numeric(net_info[[j]]$reference$year)
doi_mng <- net_info[[j]]$reference$doi
references[j,1] <- dataset_id
references[j,2] <- first_author
references[j,3] <- year_mng
references[j,4] <- doi_mng
references <- references[order(references$`Dataset ID`),]
rownames(references) <- 1:nrow(references)
}
}#End of mg refs
if(spatial==TRUE){
spatial1 <- as.data.frame(matrix(ncol = 4))
names(spatial1) <- c("Dataset ID", "first_author", "lat", "long")
message("Fetching coordinates!")
for(z in 1: length(net_info)){
dataset_id <- net_info[[z]]$dataset$dataset_id
lat_mng <- net_info[[z]]$network$geom_lat
long_mng <- net_info[[z]]$network$geom_lon
first_author <- net_info[[z]]$reference$first_author
if(length(unlist(lat_mng))>1){
spatial2 <- as.data.frame(matrix(ncol = 4))
names(spatial2) <- c("Dataset ID", "first_author", "long", "lat" )
for(b in 1:length(unlist(lat_mng))){
spatial2[b,3] <- long_mng[[1]] [b]
spatial2[b,4] <- lat_mng [[1]] [b]
}
spatial2[,1] <- dataset_id
spatial2[,2] <- first_author
spatial1 <- rbind(spatial1, spatial2)
}
spatial1[z,1] <- dataset_id
spatial1[z,2] <- first_author
if(length(unlist(lat_mng))==1) spatial1[z,3] <- lat_mng
if(length(unlist(lat_mng))==1) spatial1[z,4] <- long_mng
}
spatial1 <- spatial1[order(spatial1$`Dataset ID`),]
rownames(spatial1) <- 1:nrow(spatial1)
}#End of mg spatial
if (exists("references") & exists("spatial1")) (if(nrow(references)!=nrow(spatial1)) message("WARNING: There are more than on FW in some datasets! References and Spatial data frames have different number of rows."))
}#end of mangal
message(paste0("DONE! \n\nOverall the list stores ", length(fwlist), " datasets!"))
master_list <- list()
master_list[["int_matrix"]] <- fwlist
if(ecosyst==TRUE) {
master_list[["ecosystem"]] <- ecosystem
message ("\n Additional element in the results: \n\n The vector with information on the ecosystems.")
}
if(ref==TRUE) {
master_list[["references"]] <- references
message ("Additional element in the results! \nA data frame with information on the references.")
}
if(spatial==TRUE) {
master_list[["spatial_info"]] <- spatial1
message ("\n Additional element in the results: \n\n Spatial information was added.")
}
if(code==TRUE) {
if(db == "gw") master_list[["code"]] <- files_gw
if(db == "wl") master_list[["code"]] <- files_wl
if(db == "mg") master_list[["code"]] <- references[1,]
message ("Added food web code information.")
}
#Return results
if(length(master_list)==1) return(fwlist)
if(length(master_list)!=1) return(master_list)
message("####################### DONE! #######################")
}#END OF FUNCTION create.fw.list
To leave a comment for the author, please follow the link and comment on their blog: R Code – Geekcologist.
R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.