tests/example-GMPD/data/load_GMPD.R

In melmasri/HPprediction: Host-parasite prediction based on phylogeny

#!~/bin/R
obj.names = ls()
library(reshape2)

### Download and load data
if(!file.exists("data/GMPD_main.csv")){
	require(fulltext)
	# GMPD v2.0
	# Stephens et al. 2016 Ecology
	dt <- read.csv(unzip(ft_get_si("10.1002/ecy.1799", 1, "wiley"), exdir="data/", "GMPD_datafiles/GMPD_main.csv", junkpaths=TRUE), as.is=TRUE)
} else dt <- read.csv("data/GMPD_main.csv", as.is=TRUE)

### Formating the dataset
# Subsetting to Carnivore + Ungulate subsections
dt <- dt[dt$Group %in% c("carnivores","ungulates"),]

# Removing parasites not reported to species
Sys.setlocale('LC_ALL','C') 
dt <- dt[grep("sp[.]",dt$ParasiteCorrectedName, invert=TRUE),]
dt <- dt[grep("ABOLISHED",dt$ParasiteCorrectedName, invert=TRUE),]
dt <- dt[grep("no binomial name",dt$ParasiteCorrectedName, invert=TRUE),]
dt <- dt[grep("not identified to genus",dt$ParasiteCorrectedName, invert=TRUE),]
dt <- dt[grep("SPLITTED in ICTV",dt$ParasiteCorrectedName, invert=TRUE),]
dt <- dt[grep("Diphyllobothrium sp",dt$ParasiteCorrectedName, invert=TRUE),]

# Removing entires with "no binomial name" for hosts
dt <- dt[dt$HostCorrectedName!="no binomial name",]

# Formatting host names to match phylogeny
dt$HostCorrectedName <- gsub(" ","_", dt$HostCorrectedName)

# Remove entries with prevalence = 0 : because ParasiteCorrectedName was tested and not found.
dt <- dt[which(dt$Prevalence!=0),]


### Creating interaction matrix
# Removing spaces and other non-unique characters in Citation
Citation <- gsub('(et al|et al.| |-)', '', dt$Citation)

# Calculating unique citation count
hp <- unique(data.frame( dt[, c('HostCorrectedName', 'ParasiteCorrectedName')], Citation=Citation))

if(COUNT | length(grep('COUNT', ls()))==0){
    ## For count data
    aux <- table(paste0(hp$HostCorrectedName, hp$ParasiteCorrectedName))
    hp <- unique(hp [, c("HostCorrectedName", "ParasiteCorrectedName")])
    CiteCount <- sapply(paste0(hp$HostCorrectedName, hp$ParasiteCorrectedName), function(r){
        aux[which(names(aux)==r)]
    })
    hp <- cbind(hp, CiteCount)
    ## Creating interaction / community matrix
    com <- dcast(hp,HostCorrectedName ~ ParasiteCorrectedName,
                 value.var='CiteCount', fill=0)
}else{
## First citation years
    Citation.min.year = as.numeric(substr(hp$Citation,
        regexpr('[0-9]{2,4}',hp$Citation),regexpr('[0-9]{2,4}',hp$Citation)+3))
    
    ## Creating interaction / community matrix
    com = dcast(cbind(hp, Citation.min.year), HostCorrectedName ~ ParasiteCorrectedName,
        fun.aggregate = min, value.var = 'Citation.min.year', fill=0)
}
 
rownames(com) <- com$HostCorrectedName
com <- subset(com, select = -HostCorrectedName)
com <- as.matrix(com)

### Removing all unnecessary variables
rm(list=grep('com', setdiff(ls(), obj.names), invert=TRUE, value=T))