## R version 3.5 or later
if (!requireNamespace("BiocManager", quietly = TRUE))
  install.packages("BiocManager")

if("biomaRt" %in% rownames(installed.packages()) == FALSE) {
  BiocManager::install('biomaRt')
}
if("jsonlite" %in% rownames(installed.packages()) == FALSE) {
  BiocManager::install('jsonlite')
}
if("httr" %in% rownames(installed.packages()) == FALSE) {
  BiocManager::install('httr')
}
library(biomaRt)
library(jsonlite)
library(httr)

# PAGERCOV_ANALYSIS is a function connected to PAGER api to perform hypergeometric test and retrieve enriched PAGs associated to a list of genes
# The input parameters are:
# 1.dir: the directory of output file.
# 2.filename: the name of the output file.
# 3.markers: a list of gene symbols.
# 4.type: a list of PAG types consisting of 'P', 'A' or 'G'
# 5.minSize: the allowed minimum size of PAG genes
# 6.maxSize: the allowed maximum size of PAG genes
# 7.similarity: the similarity score cutoff
# 8.overlap: the allowed minimum overlap genes
# 9.nCoCo: the minimum nCoCo score
# 10.pValue: p-value cutoff
# 11.FDR: false discovery rate 
# 12.source: a list of sources refered to 'http://discovery.informatics.uab.edu/PAGER-COV/index.php/pages/help'

PAGERCOV_ANALYSIS = function(dir = './', fileName = 'test.txt', markers = c('BRCA1','BRCA2'), type = c("P","A","G"),
                             minSize = 2, maxSize = 5000, similarity = 0.05, overlap = 1, nCoCo = 0, pValue = 0.05,
                             FDR = 0.05, Species = 'All', source = c('PubChem','PAGER-MSigDB','PAGER-GOA','PAGER-GOA_AGG',
                                                                     'PAGER-GAD','PAGER-Pfam','PAGER-GeneSigDB','PAGER-Protein Lounge',
                                                                     'PAGER-Spike','PAGER-PheWAS','PAGER-PharmGKB','PAGER-Reactome',
                                                                     'PAGER-BioCarta','PAGER-GWAS Catalog','PAGER-KEGG','PAGER-NGS Catalog',
                                                                     'PAGER-DSigDB','PAGER-GTEx','PAGER-NCI-Nature Curated','PAGER-WikiPathway',
                                                                     'Am J Respir Crit Care Med','Microbiology and Molecular Biology Reviews',
                                                                     'Zenodo','Mouse Genome Informatics Database','Nature Cell Discovery',
                                                                     'GenBank/UniProt','Cell','The Annual Review of Cell and Developmental Biology',
                                                                     'Nature','Drugbank','Nature Medicine','Cell Host and Microbe','bioRxiv'
                             )){
  url = "http://discovery.informatics.uab.edu/PAGER-COV/index.php/geneset/pagerapi"
  marker_str = NULL
  marker_str = paste0(as.matrix(markers),collapse = "%20")
  source_str = NULL
  source_str = paste0(as.matrix(source),collapse = "%20")
  contnt <- list(
    genes = marker_str,
    type = type,
    ge = minSize,
    le = maxSize,
    sim = similarity,
    olap = overlap,
    cohesion = nCoCo,
    pvalue = pValue,
    FDR = FDR,
    organism=Species,
    source=source_str
  )
  req <- httr::POST(url,
                    body=contnt
  );
  json <- httr::content(req, as = "text")
  result = fromJSON(json)
  write.table(result,paste0(dir,fileName), row.names = F, quote =F,col.names = T,sep="\t")
  return(result)
}