Mercurial > repos > tomnl > assess_purity_msms

annotate spectral_matching.R @ 0:b6b051e7a5ef 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 f9591483bef88aef71a9f0cc0ffc75cf6eab480c
author tomnl
date Mon, 05 Mar 2018 10:08:14 -0500
parents
children 146699c00d38
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
1 library(msPurity)
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
2 library(optparse)
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
3
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
4 option_list <- list(
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
5 make_option(c("-o", "--out_dir"), type="character"),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
6 make_option("--target_db_pth", type="character"),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
7 make_option("--library_db_pth", type="character"),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
8 make_option("--ra_thres_l", default=0),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
9 make_option("--ra_thres_t", default=2),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
10 make_option("--cores", default=1),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
11 make_option("--pol", default='positive'),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
12 make_option("--ppm_tol_prod", default=10),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
13 make_option("--ppm_tol_prec", default=5),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
14 make_option("--score_thres", default=0.6),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
15 make_option("--instrument_types", type='character'),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
16 make_option("--library_sources", type='character'),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
17 make_option("--scan_ids", default=NA),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
18 make_option("--topn", default=NA),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
19 make_option("--mzML_files", type="character"),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
20 make_option("--galaxy_names", type="character"),
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
21 make_option("--create_new_database", action="store_true")
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
22
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
23 )
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
24
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
25 # store options
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
26 opt<- parse_args(OptionParser(option_list=option_list))
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
27
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
28
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
29
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
30 if (!is.null(opt$create_new_database)){
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
31 target_db_pth <- file.path(opt$out_dir, 'db_with_spectral_matching.sqlite')
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
32 file.copy(opt$target_db_pth, target_db_pth)
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
33 }else{
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
34 target_db_pth <- opt$target_db_pth
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
35 }
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
36
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
37
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
38 if (opt$instrument_types=='None'){
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
39 instrument_types <- NA
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
40 }else{
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
41 instrument_types <- trimws(strsplit(opt$instrument_types, ',')[[1]])
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
42 }
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
43 if (opt$library_sources=='None'){
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
44 library_sources <- NA
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
45 }else{
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
46 library_sources <- trimws(strsplit(opt$library_sources, ',')[[1]])
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
47 }
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
48
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
49
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
50 if (!is.na(opt$scan_ids)){
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
51 scan_ids <- trimws(strsplit(opt$scan_ids, ',')[[1]])
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
52 scan_ids <- scan_ids[scan_ids != ""]
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
53 }else{
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
54 scan_ids <- NA
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
55 }
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
56
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
57
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
58 print(instrument_types)
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
59 print(library_sources)
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
60 print(scan_ids)
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
61
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
62 result <- msPurity::spectral_matching(
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
63 target_db_pth =target_db_pth ,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
64 library_db_pth = opt$library_db_pth,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
65 ra_thres_l = opt$ra_thres_l,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
66 ra_thres_t = opt$ra_thres_t,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
67 cores = opt$cores,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
68 pol = opt$pol,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
69 ppm_tol_prod = opt$ppm_tol_prod,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
70 ppm_tol_prec = opt$ppm_tol_prec,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
71 score_thres = opt$score_thres,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
72 out_dir = opt$out_dir,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
73 topn = opt$topn,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
74 grp_peaklist = NA,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
75
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
76 instrument_types = instrument_types,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
77 library_sources = library_sources,
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
78 scan_ids = scan_ids)
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
79
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
80 print(file.path(result$result_db_pth))
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
81
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
82 write.table(result$xcms_summary_df, file.path(opt$out_dir, 'xcms_hits.tsv'), row.names=FALSE, sep='\t')
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
83
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
84 con <- DBI::dbConnect(RSQLite::SQLite(), file.path(result$result_db_pth))
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
85 # con <- DBI::dbConnect(RSQLite::SQLite(), file.path(opt$out_dir, 'result.sqlite'))
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
86
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
87 cmd <- paste('SELECT * FROM matches
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
88 LEFT JOIN library_meta ON matches.lid=library_meta.lid
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
89 LEFT JOIN s_peak_meta ON matches.pid=s_peak_meta.pid
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
90 LEFT JOIN fileinfo ON s_peak_meta.fileid=fileinfo.fileid
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
91 WHERE matches.score >= ', opt$score_thres)
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
92 print(cmd)
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
93 scan_hits <- DBI::dbGetQuery(con, cmd)
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
94
b6b051e7a5ef planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit f9591483bef88aef71a9f0cc0ffc75cf6eab480c
tomnl
parents:
diff changeset
95 write.table(scan_hits, file.path(opt$out_dir, 'scan_hits.tsv'), row.names=FALSE, sep='\t')