Mercurial > repos > dlalgroup > pmids_to_pubtator_matrix

annotate abstracts_by_pmids.R @ 0:3f4adc85ba5d 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 02:01:50 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
1 #!/usr/bin/env Rscript
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
2 #TOOL2 abstracts_by_pmids
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
3 #
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
4 #This tool retrieves for all PMIDs in each row of a table the according abstracts and saves them in additional columns.
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
5 #
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
6 #Input: Tab-delimited table with columns containing PMIDs. The names of the PMID columns should start with “PMID”, e.g. “PMID_1”, “PMID_2” etc.
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
7 #
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
8 #Output: Input table with additional columns containing abstracts corresponding to the PMIDs from PubMed.
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
9 #The abstract columns are called "ABSTRACT_1", "ABSTARCT_2" etc.
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
10 #
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
11 # Usage: $ T2_abstracts_by_pmid.R [-h] [-i INPUT] [-o OUTPUT]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
12 #
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
13 # optional arguments:
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
14 # -h, --help show help message
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
15 # -i INPUT, --input INPUT input file name. add path if file is not in working directory
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
16 # -o OUTPUT, --output OUTPUT output file name. [default "T2_output"]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
17
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
18
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
19 if ( '--install_packages' %in% commandArgs()) {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
20 print('Installing packages')
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
21 if (!require('argparse')) install.packages('argparse', repo="http://cran.rstudio.com/");
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
22 if (!require("reutils")) install.packages("reutils", repo="http://cran.rstudio.com/");
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
23 if (!require('easyPubMed')) install.packages('easyPubMed', repo="http://cran.rstudio.com/" );
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
24 if (!require('textclean')) install.packages('textclean', repo="http://cran.rstudio.com/");
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
25 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
26
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
27 suppressPackageStartupMessages(library("argparse"))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
28 library("reutils")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
29 suppressPackageStartupMessages(library("easyPubMed"))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
30 suppressPackageStartupMessages(library("textclean"))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
31
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
32 parser <- ArgumentParser()
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
33 parser$add_argument("-i", "--input",
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
34 help = "input fie name. add path if file is not in workind directory")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
35 parser$add_argument("-o", "--output", default="abstracts_by_pmids_output",
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
36 help = "output file name. [default \"%(default)s\"]")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
37 parser$add_argument("--install_packages", action="store_true", default=FALSE,
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
38 help="If you want to auto install missing required packages.")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
39
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
40 args <- parser$parse_args()
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
41
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
42 data = read.delim(args$input, stringsAsFactors=FALSE, header= TRUE, sep='\t')
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
43 pmids_cols_index <- grep("PMID", names(data))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
44
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
45 fetch_abstracts = function(PMIDs, row){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
46
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
47 efetch_result <- NULL
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
48 try_num <- 1
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
49 t_0 <- Sys.time()
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
50
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
51 while(is.null(efetch_result)) {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
52
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
53 # Timing check: kill at 3 min
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
54 if (try_num > 1){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
55 Sys.sleep(time = 1*try_num)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
56 cat("Problem to receive PubMed data or error is received. Please wait. Try number: ",try_num,"\n")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
57 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
58
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
59 t_1 <- Sys.time()
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
60
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
61 if(as.numeric(difftime(t_1, t_0, units = "mins")) > 3){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
62 message("Killing the request! Something is not working. Please, try again later","\n")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
63 return(data)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
64 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
65
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
66 efetch_result <- tryCatch({
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
67 suppressWarnings(efetch(uid=PMIDs, db="pubmed", retmode = "xml"))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
68 }, error = function(e) {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
69 NULL
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
70 })
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
71
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
72 if(!is.null(as.list(efetch_result$errors)$error)){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
73 if (as.list(efetch_result$errors)$error == "HTTP error: Status 400; Bad Request") {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
74 efetch_result <- NULL
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
75 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
76 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
77
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
78 try_num <- try_num + 1
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
79
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
80 } #while loop end
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
81
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
82 # articles to list
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
83 xml_data <- strsplit(efetch_result$content, "<PubmedArticle(>|[[:space:]]+?.*>)")[[1]][-1]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
84 xml_data <- sapply(xml_data, function(x) {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
85 #trim extra stuff at the end of the record
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
86 if (!grepl("</PubmedArticle>$", x))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
87 x <- sub("(^.*</PubmedArticle>).*$", "\\1", x)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
88 # Rebuid XML structure and proceed
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
89 x <- paste("<PubmedArticle>", x)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
90 gsub("[[:space:]]{2,}", " ", x)},
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
91 USE.NAMES = FALSE, simplify = TRUE)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
92
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
93 abstract.text = sapply(xml_data, function(x){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
94 custom_grep(x, tag="AbstractText", format="char")},
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
95 USE.NAMES = FALSE, simplify = TRUE)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
96
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
97 abstracts <- sapply(abstract.text, function(x){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
98 if (length(x) > 1){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
99 x <- paste(x, collapse = " ", sep = " ")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
100 x <- gsub("</{0,1}i>", "", x, ignore.case = T)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
101 x <- gsub("</{0,1}b>", "", x, ignore.case = T)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
102 x <- gsub("</{0,1}sub>", "", x, ignore.case = T)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
103 x <- gsub("</{0,1}exp>", "", x, ignore.case = T)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
104 } else if (length(x) < 1) {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
105 x <- NA
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
106 } else {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
107 x <- gsub("</{0,1}i>", "", x, ignore.case = T)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
108 x <- gsub("</{0,1}b>", "", x, ignore.case = T)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
109 x <- gsub("</{0,1}sub>", "", x, ignore.case = T)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
110 x <- gsub("</{0,1}exp>", "", x, ignore.case = T)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
111 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
112 x
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
113 },
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
114 USE.NAMES = FALSE, simplify = TRUE)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
115
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
116 abstracts = as.character(abstracts)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
117
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
118 if(length(abstracts)>0){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
119 data[row,sapply(1:length(abstracts),function(i){paste0("ABSTRACT_",i)})] <- abstracts
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
120 cat(length(abstracts)," abstracts for PMIDs of row ", row, " are added in the table.","\n")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
121 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
122
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
123 return(data)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
124 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
125
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
126
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
127 for(row in 1:nrow(data)){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
128 PMIDs= as.character(unique(data[row, pmids_cols_index]))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
129 PMIDs = PMIDs[!PMIDs=="NA"]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
130
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
131 if(length(PMIDs) > 0){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
132 data = tryCatch(fetch_abstracts(PMIDs, row),
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
133 error=function(e){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
134 Sys.sleep(3)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
135 })
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
136 } else {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
137 print(paste("No PMIDs in row", row))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
138 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
139 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
140
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
141 write.table(data, args$output, sep = '\t', row.names = FALSE, col.names = TRUE)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
142