text_to_wordmatrix.R

#!/usr/bin/env Rscript
# tool: text_to_wordmatrix
#
#The tool extracts the most frequent words per entity (per row). Text of columns starting with "ABSTRACT" or "TEXT" are considered.
#All extracted terms are used to generate a word matrix with rows = entities and columns = extracted words.
#The resulting matrix is binary with 0= word not present in abstracts of entity and 1= word present in abstracts of entity.
#
#Input: Output of "pubmed_by_queries" or "abstracts_by_pmids", or tab-delimited table with entities in column called “ID_<name>”,
#e.g. “ID_genes” and text in columns starting with "ABSTRACT" or "TEXT".
#
#Output: Binary matrix with rows = entities and columns = extracted words.
#
#usage: text_to_wordmatrix.R [-h] [-i INPUT] [-o OUTPUT] [-n NUMBER] [-r] [-l] [-w] [-s] [-p]
#
# optional arguments:
# -h, --help                    show help message
# -i INPUT, --input INPUT       input file name. add path if file is not in working directory
# -o OUTPUT, --output OUTPUT    output file name. [default "text_to_wordmatrix_output"]
# -n NUMBER, --number NUMBER    number of most frequent words that should be extracted [default "50"]
# -r, --remove_num              remove any numbers in text
# -l, --lower_case              by default all characters are translated to lower case. otherwise use -l
# -w, --remove_stopwords        by default a set of english stopwords (e.g., "the" or "not") are removed. otherwise use -w
# -s, --stemDoc                 apply Porter"s stemming algorithm: collapsing words to a common root to aid comparison of vocabulary
# -p, --plurals                 by default words in plural and singular are merged to the singular form. otherwise use -p

if ("--install_packages" %in% commandArgs()) {
  print("Installing packages")
  if (!require("argparse")) install.packages("argparse", repo = "http://cran.rstudio.com/");
  if (!require("PubMedWordcloud")) install.packages("PubMedWordcloud", repo = "http://cran.rstudio.com/");
  if (!require("SnowballC")) install.packages("SnowballC", repo = "http://cran.rstudio.com/");
  if (!require("textclean")) install.packages("textclean", repo = "http://cran.rstudio.com/");
  if (!require("SemNetCleaner")) install.packages("SemNetCleaner", repo = "http://cran.rstudio.com/");
  if (!require("stringi")) install.packages("stringi", repo = "http://cran.rstudio.com/");
  if (!require("stringr")) install.packages("stringr", repo = "http://cran.rstudio.com/");
}

suppressPackageStartupMessages(library("argparse"))
suppressPackageStartupMessages(library("PubMedWordcloud"))
suppressPackageStartupMessages(library("SnowballC"))
suppressPackageStartupMessages(library("SemNetCleaner"))
suppressPackageStartupMessages(library("textclean"))
suppressPackageStartupMessages(library("stringi"))
suppressPackageStartupMessages(library("stringr"))

parser <- ArgumentParser()
parser$add_argument("-i", "--input",
                    help = "input fie name. add path if file is not in workind directory")
parser$add_argument("-o", "--output", default = "text_to_wordmatrix_output",
                    help = "output file name. [default \"%(default)s\"]")
parser$add_argument("-n", "--number", type = "integer", default = 50, choices = seq(1, 500), metavar = "{0..500}",
                    help = "number of most frequent words used per ID in word matrix [default \"%(default)s\"]")
parser$add_argument("-r", "--remove_num", action = "store_true", default = FALSE,
                    help = "remove any numbers in text")
parser$add_argument("-l", "--lower_case", action = "store_false", default = TRUE,
                    help = "by default all characters are translated to lower case. otherwise use -l")
parser$add_argument("-w", "--remove_stopwords", action = "store_false", default = TRUE,
                    help = "by default a set of English stopwords (e.g., 'the' or 'not') are removed. otherwise use -s")
parser$add_argument("-s", "--stemDoc", action = "store_true", default = FALSE,
                    help = "apply Porter's stemming algorithm: collapsing words to a common root to aid comparison of vocabulary")
parser$add_argument("-p", "--plurals", action = "store_false", default = TRUE,
                    help = "by default words in plural and singular are merged to the singular form. otherwise use -p")
parser$add_argument("--install_packages", action = "store_true", default = FALSE,
                    help = "If you want to auto install missing required packages.")

args <- parser$parse_args()


data <- read.delim(args$input, stringsAsFactors = FALSE, header = TRUE, sep = "\t")
word_matrix <- data.frame()

text_cols_index <- grep(c("ABSTRACT|TEXT"), names(data))

for (row in seq(nrow(data))) {
    top_words <- cleanAbstracts(abstracts = data[row, text_cols_index],
                               rmNum = args$remove_num,
                               tolw = args$lower_case,
                               rmWords = args$remove_stopwords,
                               stemDoc = args$stemDoc)

    top_words$word <- as.character(top_words$word)

    cat("Most frequent words for row", row, " are extracted.", "\n")

    if (args$plurals == TRUE) {
      top_words$word <- sapply(top_words$word, function(x) {
        singularize(x)
        })
      top_words <- aggregate(freq~word, top_words, sum)
    }

    top_words <- top_words[order(top_words$freq, decreasing = TRUE), ]
    top_words$word <- as.character(top_words$word)

    number_extract <- min(args$number, nrow(top_words))
    word_matrix[row, sapply(1:number_extract, function(x) {
      paste0(top_words$word[x])
      })] <- top_words$freq[1:number_extract]
  }

  word_matrix <- as.matrix(word_matrix)
  word_matrix[is.na(word_matrix)] <- 0
  word_matrix <- (word_matrix > 0) * 1  #binary matrix

cat("A matrix with ", nrow(word_matrix), " rows and ", ncol(word_matrix), "columns is generated.", "\n")

write.table(word_matrix, args$output, row.names = FALSE, sep = "\t", quote = FALSE)