R/pubmedQuery.R

In andymckenzie/bayesbio: Miscellaneous Functions for Bioinformatics and Bayesian Statistics

#' @title Perform PubMed queries on 2x2 combinations of term vectors.
#' @description Perform PubMed queries on the intersections of two character vectors. This function is a wrapper to RISmed::EUtilsSummary with type = 'esearch', db = 'pubmed'.
#' @param rowTerms Character vector of terms that should make up the rows of the resulting mention count data frame.
#' @param colTerms Character vector of terms for the columns.
#' @param sleepTime How much time (in seconds) to sleep between successive PubMed queries. If you set this too low, PubMed may shut down your connection to prevent overloading their servers.
#' @param ... Additional arguments to RISmed::EUtilsSummary
#' @param use Package to use to search PubMed. Options = "rentrez", "RISmed".
#' @return A data frame of the number of mentions for each combination of terms.
#' @export
pubmedQuery <- function(rowTerms, colTerms, sleepTime = 0.01, use = "rentrez", ...){

  if(use == "RISmed"){
    if (!requireNamespace("RISmed", quietly=TRUE)) stop("Please install package 'RISmed' to use this function.")
  }
  if(use == "rentrez"){
    if (!requireNamespace("rentrez", quietly=TRUE)) stop("Please install package 'rentrez' to use this function.")
  }

  disease_gene_mentions = data.frame(matrix(0, nrow = length(rowTerms),
  	 ncol = length(colTerms) + 1))

  if(any(duplicated(rowTerms))){
    message("Duplicated entries in rowTerms; collapsing to unique terms.")
    rowTerms = unique(rowTerms)
  }

  if(any(duplicated(colTerms))){
    message("Duplicated entries in colTerms; collapsing to unique terms.")
    colTerms = unique(colTerms)
  }

  for(i in 1:length(colTerms)){
  	for(j in 1:length(rowTerms)){
      query = paste(colTerms[i], "AND", rowTerms[j], sep = " ")
      str(query)
      if(use == "RISmed"){
        res = RISmed::EUtilsSummary(query,
          type = 'esearch', db = 'pubmed', ...)
    		disease_gene_mentions[j, i] = RISmed::QueryCount(res)
    		Sys.sleep(sleepTime)
      }
      if(use == "rentrez"){
        res = entrez_search(db="pubmed", term = query)
        disease_gene_mentions[j, i] = as.numeric(res$count)
        Sys.sleep(sleepTime)
      }
  	}
  }

  total_res = numeric(length(rowTerms))
  for(j in 1:length(rowTerms)){
    if(use == "RISmed"){
      res = RISmed::EUtilsSummary(rowTerms[j], type = 'esearch', db = 'pubmed')
      total_res[j] = RISmed::QueryCount(res)
    }
    if(use == "rentrez"){
      res = entrez_search(db="pubmed", term = rowTerms[j])
      total_res[j] = as.numeric(res$count)
    }
    Sys.sleep(sleepTime)
  }

  rownames(disease_gene_mentions) = rowTerms
  disease_gene_mentions[, length(colTerms) + 1] = total_res
  colnames(disease_gene_mentions) = c(colTerms, "Total_Mentions")
  disease_gene_mentions = disease_gene_mentions[order(disease_gene_mentions[ , length(colTerms) + 1]), ]

  return(disease_gene_mentions)

}