Mercurial > repos > tomnl > dma_filelist_generation
annotate dma-filelist-generation.py @ 10:7c54e2862595 draft default tip
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 14435cfc042911bf1ee409f5a0d5ef908f5feec0-dirty
| author | tomnl | 
|---|---|
| date | Thu, 21 Jun 2018 08:45:03 -0400 | 
| parents | 6a2bb42acfe4 | 
| children | 
| rev | line source | 
|---|---|
| 0 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 1 import argparse | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 2 import textwrap | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 3 import os | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 4 import re | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 5 import collections | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 6 import csv | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 7 from operator import itemgetter | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 8 | 
| 
6a2bb42acfe4
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): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 10 filelist = [] | 
| 
6a2bb42acfe4
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): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 12 | 
| 
6a2bb42acfe4
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): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 14 | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 16 fnl = fn.split("_") | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 17 filename = fn | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 19 wells = fnl[0] | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 20 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 21 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 22 | 
| 
6a2bb42acfe4
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): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 24 polarity = "pos" | 
| 
6a2bb42acfe4
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): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 26 polarity = "neg" | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 27 else: | 
| 
6a2bb42acfe4
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" | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 29 print error_message | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 30 return 1, error_message, "", "" | 
| 
6a2bb42acfe4
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): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 32 sample_type = "blank" | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 33 else: | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 34 sample_type = "sample" | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 35 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 36 | 
| 
6a2bb42acfe4
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]) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 38 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 39 return filelist | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 40 | 
| 
6a2bb42acfe4
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): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 42 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 43 filedict = collections.OrderedDict() | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 44 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 45 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 46 if file_type=="mzML": | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 47 reg_string = "^.*mzML$" | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 48 else: | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 49 reg_string = "^.*raw$" | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 50 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 51 if isinstance(filesin, list): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 52 filelist = [] | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 53 for f in filesin: | 
| 
6a2bb42acfe4
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)) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 55 else: | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 57 | 
| 
6a2bb42acfe4
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 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 59 for f1 in filelist: | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 60 well = f1[0] | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 61 filedict[well] = [] | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 62 for f2 in filelist: | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 63 if well == f2[0]: | 
| 
6a2bb42acfe4
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]) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 65 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 66 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 67 | 
| 
6a2bb42acfe4
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)) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 69 | 
| 
6a2bb42acfe4
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(): | 
| 
6a2bb42acfe4
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] | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 72 classes.sort() | 
| 
6a2bb42acfe4
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']: | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 75 print error_message | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 76 return 1, error_message, "", "" | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 77 | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 79 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 80 if create_filelist: | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 81 | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 84 | 
| 
6a2bb42acfe4
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 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 86 | 
| 
6a2bb42acfe4
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'): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 88 | 
| 
6a2bb42acfe4
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)) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 90 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 91 if file_spacing=='tsv': | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 92 delim = '\t' | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 93 elif file_spacing=='csv': | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 94 delim = ',' | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 95 else: | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 96 delim = ',' | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 97 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 98 | 
| 
6a2bb42acfe4
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: | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 101 | 
| 
6a2bb42acfe4
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']) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 103 c = 1 | 
| 
6a2bb42acfe4
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(): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 105 for i in v: | 
| 
6a2bb42acfe4
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]) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 107 c +=1 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 108 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 109 return outname | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 110 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 111 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 112 def main(): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 113 | 
| 
6a2bb42acfe4
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', | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 115 formatter_class=argparse.RawDescriptionHelpFormatter, | 
| 
6a2bb42acfe4
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''', | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 117 epilog=textwrap.dedent(''' | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 118 ------------------------------------------------------------------------- | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 119 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 120 Example Usage | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 121 | 
| 
6a2bb42acfe4
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 . | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 123 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 124 ''')) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 125 | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
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') | 
| 
6a2bb42acfe4
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') | 
| 
6a2bb42acfe4
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') | 
| 
6a2bb42acfe4
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') | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 132 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 133 args = p.parse_args() | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 134 | 
| 
6a2bb42acfe4
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): | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 136 os.makedirs(args.o) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 137 print args.o | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 138 | 
| 
6a2bb42acfe4
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: | 
| 
6a2bb42acfe4
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' | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 141 exit() | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 142 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 143 if args.check_mzml: | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 145 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 146 if args.check_raw: | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 147 | 
| 
6a2bb42acfe4
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) | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 149 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 150 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 151 | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 152 if __name__ == '__main__': | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 153 main() | 
| 
6a2bb42acfe4
planemo upload for repository https://github.com/computational-metabolomics/dma-tools-galaxy commit 6c48bd51987a28401de6cf5e49b1b30e5e73fe16-dirty
 tomnl parents: diff
changeset | 154 | 
