Mercurial > repos > dlalgroup > pmids_to_pubtator_matrix

annotate pmids_to_pubtator_matrix.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 #tool: pmids_to_pubtator_matrix
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 #The tool uses all PMIDs per row and extracts "Gene", "Disease", "Mutation", "Chemical" and "Species" terms of the
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
5 #corresponding abstracts, using PubTator annotations. The user can choose from which categories terms should be extracted.
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
6 #The extracted terms are united in one large binary matrix, with 0= term not present in abstracts of that row and 1= term
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
7 #present in abstracts of that row. The user can decide if the extracted scientific terms should be extracted and used as
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
8 #they are or if they should be grouped by their geneIDs/ meshIDs (several terms can often be grouped into one ID).
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
9 #äAlso, by default all terms are extracted, otherwise the user can specify a number of most frequent words to be extracted per row.
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 #Input: Output of abstracts_by_pmids or tab-delimited table with columns containing PMIDs.
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
12 #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
13 #
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
14 #Output: Binary matrix in that each column represents one of the extracted terms.
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
15 #
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
16 # usage: $ pmids_to_pubtator_matrix.R [-h] [-i INPUT] [-o OUTPUT] [-n NUMBER]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
17 # [-c {Genes,Diseases,Mutations,Chemicals,Species} [{Genes,Diseases,Mutations,Chemicals,Species} ...]]
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 # optional arguments:
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
20 # -h, --help show help message
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
21 # -i INPUT, --input INPUT input file 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
22 # -n NUMBER, --number NUMBER Number of most frequent terms/IDs to extract. By default all terms/IDs are extracted.
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
23 # -o OUTPUT, --output OUTPUT output file name. [default "pmids_to_pubtator_matrix_output"]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
24 # -c {Gene,Disease,Mutation,Chemical,Species} [{Genes,Diseases,Mutations,Chemicals,Species} ...], --categories {Gene,Disease,Mutation,Chemical,Species} [{Gene,Disease,Mutation,Chemical,Species} ...]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
25 # Pubtator categories that should be considered. [default "('Gene', 'Disease', 'Mutation','Chemical')"]
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 if ( '--install_packages' %in% commandArgs()) {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
28 print('Installing packages')
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
29 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
30 if (!require('stringr')) install.packages('stringr',repo="http://cran.rstudio.com/");
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
31 if (!require('RCurl')) install.packages('RCurl',repo="http://cran.rstudio.com/");
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
32 if (!require('stringi')) install.packages('stringi',repo="http://cran.rstudio.com/");
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
33 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
34
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
35 suppressPackageStartupMessages(library("argparse"))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
36 library('stringr')
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
37 library('stringi')
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
38 library('RCurl')
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 parser <- ArgumentParser()
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 parser$add_argument("-i", "--input",
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
43 help = "input fie name. add path if file is not in workind directory")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
44 parser$add_argument("-o", "--output", default="pmids_to_pubtator_matrix_output",
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
45 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
46 parser$add_argument("-c", "--categories", choices=c("Gene", "Disease", "Mutation", "Chemical", "Species"), nargs="+",
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
47 default= c("Gene", "Disease", "Mutation", "Chemical"),
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
48 help = "Pubtator categories that should be considered. [default \"%(default)s\"]")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
49 parser$add_argument("-b", "--byid", action="store_true", default=FALSE,
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
50 help="If you want to find common gene IDs / mesh IDs instead of scientific terms.")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
51 parser$add_argument("-n", "--number", default=NULL, type="integer",
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
52 help="Number of most frequent terms/IDs to extract. By default all terms/IDs are extracted.")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
53 parser$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
54 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
55
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
56 args <- parser$parse_args()
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
57
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
58
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
59 data = read.delim(args$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
60
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
61 pmid_cols_index <- grep(c("PMID"), names(data))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
62 word_matrix = data.frame()
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
63 dict.table = data.frame()
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
64 pmids_count <- 0
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
65 pubtator_max_ids = 100
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
66
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
67 get_pubtator_terms = function(pmids, categories){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
68
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
69 table = NULL
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
70 for (pmid_split in split(pmids, ceiling(seq_along(pmids)/pubtator_max_ids))){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
71 out.data = NULL
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
72 try_num <- 1
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
73 t_0 <- Sys.time()
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
74
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
75 while(TRUE) {
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 # Timing check: kill at 3 min
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
78 if (try_num > 1){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
79 cat("Connection problem. 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
80 Sys.sleep(time = 2*try_num)
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 try_num <- try_num + 1
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
83
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
84 t_1 <- Sys.time()
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
85
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
86 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
87 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
88 return(table)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
89 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
90 out.data <- tryCatch({
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
91 getURL(paste("https://www.ncbi.nlm.nih.gov/research/pubtator-api/publications/export/pubtator?pmids=",
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
92 paste(pmid_split, collapse=","), sep = ""))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
93 }, error = function(e) {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
94 print(e)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
95 next
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
96 }, finally = {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
97 Sys.sleep(0)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
98 })
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
99
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
100 if(!is.null(out.data)){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
101 out.data = unlist(strsplit(out.data, "\n", fixed = T))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
102
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
103 # skip first few lines, is this needed?
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
104 for (i in 3:length(out.data)) {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
105 temps = unlist(strsplit(out.data[i], "\t", fixed = T))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
106 if (length(temps) == 5) {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
107 # make 5 be 6
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
108 temps = c(temps, NA)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
109 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
110 if (length(temps) == 6) {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
111 # keep only 6
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
112 table = rbind(table, temps)
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 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
115 break
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
116 }
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 } #end while loop
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
119 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
120
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
121 index.categories = c()
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
122 categories = as.character(unlist(categories))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
123
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
124 if(ncol(table) == 6){
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 for(i in categories){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
127 tmp.index = grep(TRUE, i == as.character(table[,5]))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
128
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
129 if(length(tmp.index) > 0){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
130 index.categories = c(index.categories,tmp.index)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
131 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
132 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
133
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
134 table = as.data.frame(table, stringsAsFactors=FALSE)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
135 table = table[index.categories,c(4,6)]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
136 table = table[!is.na(table[,2]),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
137 table = table[!(table[,2] == "NA"),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
138 table = table[!(table[,1] == "NA"),]
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 if(args$byid){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
141 if(!is.null(args$number)){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
142 #retrieve top X mesh.ids
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
143 table.mesh = as.data.frame(table(table[,2]))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
144 colnames(table.mesh)[1] = "mesh.id"
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
145 table = table[order(table.mesh$Freq, decreasing = TRUE),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
146 table = table[1:min(args$number, nrow(table.mesh)),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
147 table.mesh$mesh.id = as.character(table.mesh$mesh.id)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
148 #subset table for top X mesh.ids
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
149 table = table[which(as.character(table$V6) %in% as.character(table.mesh$mesh.id)),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
150 table = table[!duplicated(table[,2]),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
151 }else{
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
152 table = table[!duplicated(table[,2]),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
153 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
154 } else {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
155 if(!is.null(args$number)){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
156 table[,1] = tolower(as.character(table[,1]))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
157 table = as.data.frame(table(table[,1]))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
158 colnames(table)[1] = "term"
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
159 table = table[order(table$Freq, decreasing = TRUE),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
160 table = table[1:min(args$number, nrow(table)),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
161 table$term = as.character(table$term)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
162
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
163 }else{
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
164 table[,1] = tolower(as.character(table[,1]))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
165 table = table[!duplicated(table[,1]),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
166 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
167 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
168
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
169 return(table)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
170
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
171 } else {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
172 return(NULL)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
173 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
174 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
175
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
176
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
177 #for all PMIDs of a row get PubTator terms and add them to the matrix
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
178 for (i in 1:nrow(data)){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
179
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
180 pmids = as.character(data[i,pmid_cols_index])
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
181 pmids = pmids[!pmids == "NA"]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
182
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
183
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
184 if ( (pmids_count > 10000)){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
185 cat("Break (10s) to avoid killing of requests. Please wait.",'\n')
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
186 Sys.sleep(10)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
187 pmids_count = 0
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
188 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
189
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
190 pmids_count = pmids_count + length(pmids)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
191
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
192 #get puptator terms with get_pubtator_terms function
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
193 if (length(pmids) >0){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
194 table = get_pubtator_terms(pmids, args$categories)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
195
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
196 if(!is.null(table)){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
197
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
198 colnames(table)= c("term","mesh.id")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
199
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
200 # add data in binary matrix
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
201 if (args$byid){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
202 mesh.ids = as.character(table$mesh.id)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
203 if (length(mesh.ids) > 0 ){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
204 word_matrix[i,mesh.ids] <- 1
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
205 cat(length(mesh.ids), " IDs for PMIDs of row", i," were added",'\n')
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
206 # add data in dictionary
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
207 dict.table = rbind(dict.table, table)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
208 dict.table = dict.table[!duplicated(as.character(dict.table[,2])),]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
209 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
210 } else {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
211 terms = as.character(table[,1])
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
212 if (length(terms) > 0 ){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
213 word_matrix[i,terms] <- 1
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
214 cat(length(terms), " terms for PMIDs of row", i," were added.",'\n')
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
215 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
216 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
217 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
218
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
219 } else {
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
220 cat("No terms for PMIDs of row", i," were found.",'\n')
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
221 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
222 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
223
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
224 if (args$byid){
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
225 #change column names of matrix: exchange mesh ids/ids with term
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
226 index_names = match(names(word_matrix), as.character(dict.table[[2]]))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
227 names(word_matrix) = dict.table[index_names,1]
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
228 }
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
229
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
230 colnames(word_matrix) = gsub("[^[:print:]]","",colnames(word_matrix))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
231 colnames(word_matrix) = gsub('\"', "", colnames(word_matrix), fixed = TRUE)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
232
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
233 #merge duplicated columns
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
234 word_matrix = as.data.frame(do.call(cbind, by(t(word_matrix),INDICES=names(word_matrix),FUN=colSums)))
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
235
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
236 #save binary matrix
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
237 word_matrix <- as.matrix(word_matrix)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
238 word_matrix[is.na(word_matrix)] <- 0
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
239 cat("Matrix with ",nrow(word_matrix)," rows and ",ncol(word_matrix)," columns generated.","\n")
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
240 #write.table(word_matrix, args$output)
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
241 write.table(word_matrix, args$output, row.names = FALSE, sep = '\t')
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
242
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
243
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
244
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
245
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
246
3f4adc85ba5d "planemo upload for repository https://github.com/dlal-group/simtext commit fd3f5b7b0506fbc460f2a281f694cb57f1c90a3c-dirty"
dlalgroup
parents:
diff changeset
247