Mercurial > repos > dlalgroup > simtext_app

comparison text_to_wordmatrix.R @ 0:34ed44f3f85c draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
author dlalgroup
date Thu, 24 Sep 2020 02:17:05 +0000
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:34ed44f3f85c
1 #!/usr/bin/env Rscript
2 # tool: text_to_wordmatrix
3 #
4 #The tool extracts the most frequent words per entity (per row). Text of columns starting with "ABSTRACT" or "TEXT" are considered.
5 #All extracted terms are used to generate a word matrix with rows = entities and columns = extracted words.
6 #The resulting matrix is binary with 0= word not present in abstracts of entity and 1= word present in abstracts of entity.
7 #
8 #Input: Output of 'pubmed_by_queries' or 'abstracts_by_pmids', or tab-delimited table with entities in column called “ID_<name>”,
9 #e.g. “ID_genes” and text in columns starting with "ABSTRACT" or "TEXT".
10 #
11 #Output: Binary matrix with rows = entities and columns = extracted words.
12 #
13 #usage: text_to_wordmatrix.R [-h] [-i INPUT] [-o OUTPUT] [-n NUMBER] [-r] [-l] [-w] [-s] [-p]
14 #
15 # optional arguments:
16 # -h, --help show help message
17 # -i INPUT, --input INPUT input file name. add path if file is not in working directory
18 # -o OUTPUT, --output OUTPUT output file name. [default "text_to_wordmatrix_output"]
19 # -n NUMBER, --number NUMBER number of most frequent words that should be extracted [default "50"]
20 # -r, --remove_num remove any numbers in text
21 # -l, --lower_case by default all characters are translated to lower case. otherwise use -l
22 # -w, --remove_stopwords by default a set of english stopwords (e.g., 'the' or 'not') are removed. otherwise use -w
23 # -s, --stemDoc apply Porter's stemming algorithm: collapsing words to a common root to aid comparison of vocabulary
24 # -p, --plurals by default words in plural and singular are merged to the singular form. otherwise use -p
25
26 if ( '--install_packages' %in% commandArgs()) {
27 print('Installing packages')
28 if (!require('argparse')) install.packages('argparse', repo="http://cran.rstudio.com/");
29 if (!require("PubMedWordcloud")) install.packages("PubMedWordcloud", repo="http://cran.rstudio.com/");
30 if (!require('SnowballC')) install.packages('SnowballC', repo="http://cran.rstudio.com/");
31 if (!require('textclean')) install.packages('textclean', repo="http://cran.rstudio.com/");
32 if (!require('SemNetCleaner')) install.packages('SemNetCleaner',repo="http://cran.rstudio.com/");
33 if (!require('stringi')) install.packages('stringi',repo="http://cran.rstudio.com/");
34 if (!require('stringr')) install.packages('stringr',repo="http://cran.rstudio.com/");
35 }
36
37 suppressPackageStartupMessages(library("argparse"))
38 suppressPackageStartupMessages(library("PubMedWordcloud"))
39 suppressPackageStartupMessages(library("SnowballC"))
40 suppressPackageStartupMessages(library("SemNetCleaner"))
41 suppressPackageStartupMessages(library("textclean"))
42 suppressPackageStartupMessages(library("stringi"))
43 suppressPackageStartupMessages(library("stringr"))
44
45 parser <- ArgumentParser()
46 parser$add_argument("-i", "--input",
47 help = "input fie name. add path if file is not in workind directory")
48 parser$add_argument("-o", "--output", default="text_to_wordmatrix_output",
49 help = "output file name. [default \"%(default)s\"]")
50 parser$add_argument("-n", "--number", type="integer", default=50, choices=seq(1, 500), metavar="{0..500}",
51 help="number of most frequent words used per ID in word matrix [default \"%(default)s\"]")
52 parser$add_argument("-r", "--remove_num", action="store_true", default=FALSE,
53 help= "remove any numbers in text")
54 parser$add_argument("-l", "--lower_case", action="store_false", default=TRUE,
55 help="by default all characters are translated to lower case. otherwise use -l")
56 parser$add_argument("-w", "--remove_stopwords", action="store_false", default=TRUE,
57 help="by default a set of English stopwords (e.g., 'the' or 'not') are removed. otherwise use -s")
58 parser$add_argument("-s", "--stemDoc", action="store_true", default=FALSE,
59 help="apply Porter's stemming algorithm: collapsing words to a common root to aid comparison of vocabulary")
60 parser$add_argument("-p", "--plurals", action="store_false", default=TRUE,
61 help="by default words in plural and singular are merged to the singular form. otherwise use -p")
62 parser$add_argument("--install_packages", action="store_true", default=FALSE,
63 help="If you want to auto install missing required packages.")
64
65 args <- parser$parse_args()
66
67
68 data = read.delim(args$input, stringsAsFactors=FALSE, header = TRUE, sep='\t')
69 word_matrix = data.frame()
70
71 text_cols_index <- grep(c("ABSTRACT|TEXT"), names(data))
72
73 for(row in 1:nrow(data)){
74 top_words = cleanAbstracts(abstracts= data[row,text_cols_index],
75 rmNum = args$remove_num,
76 tolw= args$lower_case,
77 rmWords= args$remove_stopwords,
78 stemDoc= args$stemDoc)
79
80 top_words$word <- as.character(top_words$word)
81
82 # δ γ ε
83
84 cat("Most frequent words for row", row, " are extracted.", "\n")
85
86 if(args$plurals == TRUE){
87 top_words$word <- sapply(top_words$word, function(x){singularize(x)})
88 top_words = aggregate(freq~word,top_words,sum)
89 }
90
91 top_words = top_words[order(top_words$freq, decreasing = TRUE), ]
92 top_words$word = as.character(top_words$word)
93
94 number_extract = min(args$number, nrow(top_words))
95 word_matrix[row,sapply(1:number_extract, function(x){paste0(top_words$word[x])})] <- top_words$freq[1:number_extract]
96 }
97
98 word_matrix <- as.matrix(word_matrix)
99 word_matrix[is.na(word_matrix)] <- 0
100 word_matrix <- (word_matrix>0) *1 #binary matrix
101
102 cat("A matrix with ", nrow(word_matrix), " rows and ", ncol(word_matrix), "columns is generated.", "\n")
103
104 write.table(word_matrix, args$output, row.names = FALSE, sep = '\t')