In KrishnaTO/OntologyTermAggregator: Term based Ontology parser via major repo

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

#library(OntologyTermAggregator)
# POST form request to API
library(RCurl)
# Parse XML response from API
library(XML)
# Parse JSON data
library(jsonlite)
# Data manipulation
library(tidyverse)

# Instructions on getting an API key are at http://www.bioontology.org/wiki/index.php/Annotator_User_Guide#Annotator_Web_service_Validation
#yourAPIKey <- "ENTER_API_KEY_HERE"
searchTerm <- "Neurofibroma"

# Retrieved function cat_lists: https://stackoverflow.com/a/57171812
cat_lists <- function(list1, list2) {  

  keys <- unique(c(names(list1), names(list2)))
  map2(list1[keys], list2[keys], c) %>% 
    set_names(keys)  

}

res.search <- postForm('http://data.bioontology.org/search',
                          q = searchTerm,
                          also_search_obsolete=FALSE ,
                          inlcude="all",
                          pagesize=1000,
                          apikey = yourAPIKey
                          )

res.json <- fromJSON(res.search)$collection %>% 
  filter(ontologyType == "ONTOLOGY") %>% 
  select(-obsolete, -ontologyType, -provisional, -`@context`, -cui, -semanticType)
res.json$links$`@context` <- NULL

# For convenience, as getting all axiom term data takes a while, a local copy of the results is stored in the data
if(searchTerm == "Neurofibroma" & file.exists("../data/Searchtoclasses_json.RData")){
  load("../data/Searchtoclasses_json.RData")
  } else{ 
    rdfs <- lapply(res.json$links$self, getFullEntity)
    save(rdfs, file = "../data/Searchtoclasses_json.RData")
  }

combined_output <- reduce(rdfs, cat_lists)
combined_output.parents <- reduce(combined_output$parents, cat_lists)
combined_output$parents <- NULL

cols <- c("subClassOf", "definition", "synonym", "cui", "semanticType", "label")
combined_output[cols] <- lapply(combined_output[cols], listUnique)
combined_output.parents[cols] <- lapply(combined_output.parents[cols], listunique)

# complex variables here consist of non-specific and specific relationary terms belonging to the term superclass, and thus link to input information per unique variable data input is a future work-in-process. 
complex_cols <- c("properties", "links", "@context")
combined_output.complex <- combined_output[complex_cols]
combined_output[complex_cols] <- NULL

combined_output <- lapply(combined_output, unique)
combined_output$id <- unique(combined_output$id)
combined_output$prefLabel <- unique(combined_output$prefLabel)

lapply(combined_output, function(x) write.table(data.frame(x), "combined_output.csv", append= T, sep=',' ))
saveRDS(combined_output.complex, file = "combined_output.complex.RData")