Mercurial > repos > tomnl > cameradims
annotate dma-filelist-generation.py @ 19:d4081e4160af draft
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 14435cfc042911bf1ee409f5a0d5ef908f5feec0-dirty
author | tomnl |
---|---|
date | Thu, 21 Jun 2018 08:47:44 -0400 |
parents | 55aa2dd24828 |
children |
rev | line source |
---|---|
0
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
1 import argparse |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
2 import textwrap |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
3 import os |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
4 import re |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
5 import collections |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
6 import csv |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
7 from operator import itemgetter |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
8 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
9 def check_folder(in_dir, reg_string): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
10 filelist = [] |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
11 for f in os.listdir(in_dir): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
12 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
13 if re.match(reg_string, f, re.IGNORECASE): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
14 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
15 fn, fe = os.path.splitext(f) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
16 fnl = fn.split("_") |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
17 filename = fn |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
18 fullpth = os.path.join(in_dir, f) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
19 wells = fnl[0] |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
20 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
21 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
22 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
23 if re.match("^.*pos.*", f, re.IGNORECASE): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
24 polarity = "pos" |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
25 elif re.match("^.*neg.*", f, re.IGNORECASE): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
26 polarity = "neg" |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
27 else: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
28 error_message = "Files need to have either 'pos' or 'neg' in the file name" |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
29 print error_message |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
30 return 1, error_message, "", "" |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
31 if re.match("^.*blank.*", f, re.IGNORECASE): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
32 sample_type = "blank" |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
33 else: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
34 sample_type = "sample" |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
35 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
36 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
37 filelist.append([wells, sample_type, polarity, filename, fullpth]) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
38 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
39 return filelist |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
40 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
41 def get_filelist(filesin, o, file_type='mzML', create_filelist=True): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
42 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
43 filedict = collections.OrderedDict() |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
44 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
45 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
46 if file_type=="mzML": |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
47 reg_string = "^.*mzML$" |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
48 else: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
49 reg_string = "^.*raw$" |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
50 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
51 if isinstance(filesin, list): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
52 filelist = [] |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
53 for f in filesin: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
54 filelist.extend(check_folder(f, reg_string)) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
55 else: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
56 filelist = check_folder(filesin, reg_string) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
57 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
58 # Turn filelist into a dictionary |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
59 for f1 in filelist: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
60 well = f1[0] |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
61 filedict[well] = [] |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
62 for f2 in filelist: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
63 if well == f2[0]: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
64 filedict[well].append(f2[1:len(f2)+1]) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
65 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
66 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
67 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
68 filelist = sorted(filelist, key=itemgetter(0, 1)) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
69 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
70 for k, v in filedict.iteritems(): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
71 classes = [i[0] for i in v] |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
72 classes.sort() |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
73 if not classes == ['blank', 'sample']: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
74 error_message = "!!!!ERROR!!!! Blank and sample required for each well, file type {}".format(file_type) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
75 print error_message |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
76 return 1, error_message, "", "" |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
77 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
78 print 'files of type {} checked, files OK'.format(file_type) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
79 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
80 if create_filelist: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
81 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
82 outname = write_filedict(filedict, o, file_type) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
83 print 'filelist created in folder {}, using file type {}, full path {}'.format(o, file_type, outname) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
84 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
85 return 0, "files OK", filedict, filelist |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
86 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
87 def write_filedict(filedict, out_dir, file_type, file_spacing='tsv'): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
88 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
89 outname = os.path.join(out_dir,'filelist_{}.{}'.format(file_type, file_spacing)) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
90 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
91 if file_spacing=='tsv': |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
92 delim = '\t' |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
93 elif file_spacing=='csv': |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
94 delim = ',' |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
95 else: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
96 delim = ',' |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
97 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
98 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
99 with open(outname, 'wb') as csvfile: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
100 w = csv.writer(csvfile, delimiter=delim) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
101 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
102 w.writerow(['filename','classLabel', 'multilist', 'multilistLabel']) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
103 c = 1 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
104 for k, v in filedict.iteritems(): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
105 for i in v: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
106 w.writerow([os.path.basename(i[3]), i[0], c, k]) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
107 c +=1 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
108 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
109 return outname |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
110 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
111 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
112 def main(): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
113 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
114 p = argparse.ArgumentParser(prog='PROG', |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
115 formatter_class=argparse.RawDescriptionHelpFormatter, |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
116 description='''Create filelist for DMA DIMS nearline workflow''', |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
117 epilog=textwrap.dedent(''' |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
118 ------------------------------------------------------------------------- |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
119 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
120 Example Usage |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
121 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
122 python dma-filelist-generation.py -i [dir with sample files], [dir with blank files] -o . |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
123 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
124 ''')) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
125 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
126 p.add_argument('-i', dest='i', help='dir with sample files', nargs = '*', required=True) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
127 p.add_argument('-o', dest='o', help='out dir', required=True) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
128 p.add_argument('--check_mzml', dest='check_mzml', action='store_true') |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
129 p.add_argument('--check_raw', dest='check_raw', action='store_true') |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
130 p.add_argument('--create_filelist_mzml', dest='create_filelist_mzml', action='store_true') |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
131 p.add_argument('--create_filelist_raw', dest='create_filelist_raw', action='store_true') |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
132 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
133 args = p.parse_args() |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
134 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
135 if not os.path.exists(args.o): |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
136 os.makedirs(args.o) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
137 print args.o |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
138 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
139 if not args.check_mzml and not args.check_raw: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
140 print '--check_mzml or --check_raw (or both) are required as inputs' |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
141 exit() |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
142 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
143 if args.check_mzml: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
144 get_filelist(args.i, args.o, file_type='mzML', create_filelist=args.create_filelist_mzml) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
145 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
146 if args.check_raw: |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
147 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
148 get_filelist(args.i, args.o, file_type='raw', create_filelist=args.create_filelist_raw) |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
149 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
150 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
151 |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
152 if __name__ == '__main__': |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
153 main() |
55aa2dd24828
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
tomnl
parents:
diff
changeset
|
154 |