Search NCBI using R

ncbi_search.R

library(rvest)
library(dplyr)
#browseURL(paste0(url_ncbi,qt))

qt <- "ID3" #searh query

NCBI_search <- function(qt){
  url_ncbi <- "https://www.ncbi.nlm.nih.gov/gene/?term="
  url_t <- paste0(url_ncbi,qt)
  html_t <- read_html(url_t)
  df_t <- html_table(html_t)
  df_t <- df_t[[which(sapply(df_t, function(x)any(colnames(x)=="Description")))]]
  human <- dplyr::filter(df_t, grepl("human", Description)) %>% 
    mutate(id = unlist(strsplit(`Name/Gene ID`, " "))[4])
  url_g <- paste0("https://www.ncbi.nlm.nih.gov/gene/", human$id)
  html_g <- read_html(url_g)
  txt_g <- html_element(html_g, xpath = '//*[@id="summaryDl"]') %>% 
    html_text() %>% 
    strsplit(split = "\n")
  i <- grep("Summary$",txt_g[[1]])
  res <- gsub("^ +", "", txt_g[[1]][i+1])
  return(res)  
}