Mercurial > repos > tomnl > msp_split

annotate topn.R @ 1:0db4c33f5f5b 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/dma-tools-galaxy commit af689d3f20c86f69aa824545e668280bcd5e0cca
author tomnl
date Mon, 14 May 2018 08:10:38 -0400
parents aa55ff1d76d4
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
1 library(XCMSwrapper)
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
2 library(optparse)
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
3
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
4
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
5 option_list <- list(
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
6 make_option("--sample_metadata", type="character"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
7 make_option("--out_dir", type="character"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
8 make_option("--class", type="character"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
9 make_option("--file_include", type="character"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
10 make_option("--cores", type="character"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
11 make_option("--scan_dens", type="numeric"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
12 make_option("--n", type="numeric"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
13 make_option("--ppm", type="numeric"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
14 make_option("--snthr", type="numeric"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
15 make_option("--minfrac", type="numeric"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
16 make_option("--polarity", type="character"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
17 make_option("--time_window_choice", type="character"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
18 make_option("--files", type="character"),
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
19 make_option("--galaxy_files", type="character")
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
20
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
21 )
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
22
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
23 # store options
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
24 opt<- parse_args(OptionParser(option_list=option_list))
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
25
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
26
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
27 time_window_choice<- trimws(strsplit(opt$time_window_choice, ';')[[1]])
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
28 time_window_choice <- time_window_choice[time_window_choice != ""]
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
29 tdf <- plyr::ldply(time_window_choice, function(x){trimws(strsplit(x, ',')[[1]])})
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
30 colnames(tdf) <- c('min', 'max')
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
31 tdf$min <- as.numeric(tdf$min)
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
32 tdf$max <- as.numeric(tdf$max)
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
33
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
34 # Nearline processing
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
35 params <- list(
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
36 topn.class = opt$class,
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
37 topn.class.files = opt$file_include,
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
38 topn.msPurity = TRUE,
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
39 topn.cores = opt$cores,
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
40 topn.scanDens = opt$scan_dens, # for scan range x to y, take every 'n' scan (i.e. x, x+n, x+2n ... y)
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
41 topn.timeRange = tdf,
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
42 topn.chunks = NULL, #list(c = 3, timeWin = c(0,1800)), # e.g. list(c = 3, timeWin = c(0,1800)) #chunk time range 0-1800 in to 3.
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
43 topn.n = opt$n, # number of peaks to take per window
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
44 topn.ppm = opt$ppm,
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
45 topn.snthr = opt$snthr,
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
46 topn.minfrac= opt$minfrac,
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
47 topn.clustType = 'hc',
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
48 topn.camera_xcms = '',
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
49 faahKO=FALSE
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
50 )
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
51
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
52
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
53 #
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
54 sample_metadata <- read.table(opt$sample_metadata, sep='\t', stringsAsFactors=FALSE)
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
55
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
56 filepaths <- trimws(strsplit(opt$files, ',')[[1]])
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
57 filepaths <- filepaths[filepaths != ""]
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
58 filepaths_wo_suffix <- unlist(lapply(filepaths, function(x){tools::file_path_sans_ext(basename(x))}))
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
59
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
60 galaxy_files <- trimws(strsplit(opt$galaxy_files, ',')[[1]])
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
61 galaxy_files <- galaxy_files[galaxy_files != ""]
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
62
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
63 if(!is.null(opt$class)){
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
64 # print(sample_metadata[,1])
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
65 file_select <- sample_metadata[,1][sample_metadata[,2]==opt$class]
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
66 chosen_files <- galaxy_files[filepaths_wo_suffix %in% file_select]
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
67 }else{
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
68 chosen_files <- galaxy_files
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
69 }
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
70 print(chosen_files)
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
71
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
72 if(!is.null(opt$file_include)){
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
73 file_include <- trimws(strsplit(opt$file_include, ',')[[1]])
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
74 file_include <- as.numeric(file_include[file_include != ""])
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
75 print(file_include)
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
76 chosen_files <- chosen_files[file_include]
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
77 }
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
78 print(chosen_files)
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
79 params$topn.filenames = chosen_files
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
80
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
81 topn = getTopn(obj=NULL, params)
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
82 colnames(topn) = c('mzmed', 'rtmin', 'rtmax')
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
83 topn = create_incl(grps = topn, incl_in = NULL, write_file = F, posneg = opt$polarity, comments='topn')
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
84
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
85
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
86 print('HEAD ROWS OF TOPN')
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
87 print(head(topn))
aa55ff1d76d4 planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit dcfc95273101a7ef0405c2efb8d83f5d456ccd15
tomnl
parents:
diff changeset
88 write.table(topn, file.path(opt$out_dir, 'topn_peaklist.tsv'), row.names=FALSE, sep='\t')