Mercurial > repos > tomnl > anticipated_purity_lcms

annotate spectral_matching.R @ 8:e28fc4fd51e9 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit 0aa10df0ec1ed71601f932cfb11d7d4d4f620d80-dirty
author tomnl
date Wed, 02 May 2018 13:21:00 -0400
parents b83bcc259b76
children e05068fe15a1
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
1 library(msPurity)
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
2 library(optparse)
2
b83bcc259b76 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit 2e847122cf605951c334858455fc1d3ebdb189e9-dirty
tomnl
parents: 0
diff changeset
3 print(sessionInfo())
b83bcc259b76 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit 2e847122cf605951c334858455fc1d3ebdb189e9-dirty
tomnl
parents: 0
diff changeset
4
0
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
5
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
6 option_list <- list(
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
7 make_option(c("-o", "--out_dir"), type="character"),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
8 make_option("--target_db_pth", type="character"),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
9 make_option("--library_db_pth", type="character"),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
10 make_option("--ra_thres_l", default=0),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
11 make_option("--ra_thres_t", default=2),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
12 make_option("--cores", default=1),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
13 make_option("--pol", default='positive'),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
14 make_option("--ppm_tol_prod", default=10),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
15 make_option("--ppm_tol_prec", default=5),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
16 make_option("--score_thres", default=0.6),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
17 make_option("--instrument_types", type='character'),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
18 make_option("--library_sources", type='character'),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
19 make_option("--scan_ids", default=NA),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
20 make_option("--topn", default=NA),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
21 make_option("--mzML_files", type="character"),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
22 make_option("--galaxy_names", type="character"),
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
23 make_option("--create_new_database", action="store_true")
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
24
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
25 )
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
26
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
27 # store options
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
28 opt<- parse_args(OptionParser(option_list=option_list))
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
29
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
30
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
31
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
32 if (!is.null(opt$create_new_database)){
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
33 target_db_pth <- file.path(opt$out_dir, 'db_with_spectral_matching.sqlite')
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
34 file.copy(opt$target_db_pth, target_db_pth)
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
35 }else{
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
36 target_db_pth <- opt$target_db_pth
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
37 }
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
38
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
39
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
40 if (opt$instrument_types=='None'){
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
41 instrument_types <- NA
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
42 }else{
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
43 instrument_types <- trimws(strsplit(opt$instrument_types, ',')[[1]])
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
44 }
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
45 if (opt$library_sources=='None'){
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
46 library_sources <- NA
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
47 }else{
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
48 library_sources <- trimws(strsplit(opt$library_sources, ',')[[1]])
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
49 }
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
50
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
51
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
52 if (!is.na(opt$scan_ids)){
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
53 scan_ids <- trimws(strsplit(opt$scan_ids, ',')[[1]])
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
54 scan_ids <- scan_ids[scan_ids != ""]
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
55 }else{
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
56 scan_ids <- NA
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
57 }
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
58
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
59
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
60 print(instrument_types)
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
61 print(library_sources)
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
62 print(scan_ids)
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
63
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
64 result <- msPurity::spectral_matching(
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
65 target_db_pth =target_db_pth ,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
66 library_db_pth = opt$library_db_pth,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
67 ra_thres_l = opt$ra_thres_l,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
68 ra_thres_t = opt$ra_thres_t,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
69 cores = opt$cores,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
70 pol = opt$pol,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
71 ppm_tol_prod = opt$ppm_tol_prod,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
72 ppm_tol_prec = opt$ppm_tol_prec,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
73 score_thres = opt$score_thres,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
74 out_dir = opt$out_dir,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
75 topn = opt$topn,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
76 grp_peaklist = NA,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
77
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
78 instrument_types = instrument_types,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
79 library_sources = library_sources,
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
80 scan_ids = scan_ids)
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
81
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
82 print(file.path(result$result_db_pth))
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
83
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
84 write.table(result$xcms_summary_df, file.path(opt$out_dir, 'xcms_hits.tsv'), row.names=FALSE, sep='\t')
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
85
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
86 con <- DBI::dbConnect(RSQLite::SQLite(), file.path(result$result_db_pth))
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
87 # con <- DBI::dbConnect(RSQLite::SQLite(), file.path(opt$out_dir, 'result.sqlite'))
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
88
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
89 cmd <- paste('SELECT * FROM matches
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
90 LEFT JOIN library_meta ON matches.lid=library_meta.lid
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
91 LEFT JOIN s_peak_meta ON matches.pid=s_peak_meta.pid
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
92 LEFT JOIN fileinfo ON s_peak_meta.fileid=fileinfo.fileid
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
93 WHERE matches.score >= ', opt$score_thres)
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
94 print(cmd)
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
95 scan_hits <- DBI::dbGetQuery(con, cmd)
dc00b7222a3c planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
96
2
b83bcc259b76 planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit 2e847122cf605951c334858455fc1d3ebdb189e9-dirty
tomnl
parents: 0
diff changeset
97 write.table(scan_hits, file.path(opt$out_dir, 'scan_hits.tsv'), row.names=FALSE, sep='\t')