Mercurial > repos > iuc > ena_upload

annotate process_xlsx.py @ 0:382518f24d6d draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
author iuc
date Sat, 28 Nov 2020 09:45:44 +0000
parents
children 57251c760cab
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
1 import argparse
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
2 import pathlib
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
3 import sys
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
4
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
5 import xlrd
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
6
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
7
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
8 FILE_FORMAT = 'fastq'
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
9
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
10
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
11 def extract_data(xl_sheet, expected_columns):
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
12 """
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
13 1. Check that the columns I expect are present in the sheet
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
14 (any order and mixed with others, it's just a verification that
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
15 the user filled the correct template)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
16 2. Fill a dictionary with the rows data indexed by first column in list"""
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
17 sheet_columns = {}
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
18 for sh_col in range(xl_sheet.ncols):
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
19 if xl_sheet.cell(0, sh_col).value in expected_columns:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
20 if xl_sheet.cell(0, sh_col).value in sheet_columns.keys():
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
21 sys.exit("Duplicated columns")
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
22 else:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
23 sheet_columns[xl_sheet.cell(0, sh_col).value] = sh_col
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
24 for col in range(len(expected_columns)):
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
25 assert expected_columns[col] in sheet_columns.keys(), \
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
26 "Expected column %s not found" % expected_columns[col]
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
27
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
28 # fetch rows in a dict
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
29 data_dict = {}
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
30 # the first of the expected columns will be the index
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
31 index_col = sheet_columns[expected_columns[0]]
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
32 # skip first 2 rows: column names + comments rows
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
33 for row_id in range(2, xl_sheet.nrows):
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
34 row_dict = {}
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
35 for col in range(1, len(expected_columns)):
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
36 sheet_col_index = sheet_columns[expected_columns[col]]
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
37 row_dict[expected_columns[col]] = xl_sheet.cell(row_id, sheet_col_index).value
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
38 # should check for duplicate alias/ids?
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
39 data_dict[xl_sheet.cell(row_id, index_col).value] = row_dict
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
40 return data_dict
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
41
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
42
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
43 parser = argparse.ArgumentParser()
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
44 parser.add_argument('--form', dest='xlsx_path', required=True)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
45 parser.add_argument('--out_dir', dest='out_path', required=True)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
46 parser.add_argument('--action', dest='action', required=True)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
47 parser.add_argument('--vir', dest='viral_submission', required=False, action='store_true')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
48 args = parser.parse_args()
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
49
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
50 xl_workbook = xlrd.open_workbook(args.xlsx_path)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
51
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
52 # PARSE STUDIES
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
53 #################
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
54 xl_sheet = xl_workbook.sheet_by_name('ENA_study')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
55 if xl_sheet.nrows < 3:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
56 raise ValueError('No entries found in studies sheet')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
57 studies_dict = {}
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
58 studies_col = ['alias', 'title', 'study_type', 'study_abstract']
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
59 studies_dict = extract_data(xl_sheet, studies_col)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
60
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
61 # PARSE SAMPLES
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
62 #################
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
63 xl_sheet = xl_workbook.sheet_by_name('ENA_sample')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
64 if xl_sheet.nrows < 3:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
65 raise ValueError('No entries found in samples')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
66 if args.viral_submission:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
67 samples_cols = ['alias', 'title', 'scientific_name', 'sample_description',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
68 'geographic location (country and/or sea)', 'host common name',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
69 'host health state', 'host sex', 'host scientific name', 'collector name',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
70 'collection date', 'collecting institution', 'isolate']
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
71 else:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
72 samples_cols = ['alias', 'title', 'scientific_name', 'sample_description']
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
73 samples_dict = extract_data(xl_sheet, samples_cols)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
74
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
75 # PARSE EXPERIMENTS
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
76 #################
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
77 xl_sheet = xl_workbook.sheet_by_name('ENA_experiment')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
78 if xl_sheet.nrows < 3:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
79 raise ValueError('No experiments found in experiments sheet')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
80 exp_columns = ['alias', 'title', 'study_alias', 'sample_alias', 'design_description', 'library_name',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
81 'library_strategy', 'library_source', 'library_selection', 'library_layout',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
82 'insert_size', 'library_construction_protocol', 'platform', 'instrument_model']
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
83 experiments_dict = extract_data(xl_sheet, exp_columns)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
84
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
85 # PARSE RUNS SHEET
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
86 #################
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
87 xl_sheet = xl_workbook.sheet_by_name('ENA_run')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
88 if xl_sheet.nrows < 3:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
89 raise ValueError('No entries found in runs sheet')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
90 run_cols = ['alias', 'experiment_alias', 'file_name', 'file_format']
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
91 runs_dict = extract_data(xl_sheet, run_cols)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
92
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
93 # WRITE HEADERS TO TABLES
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
94 studies_table = open(pathlib.Path(args.out_path) / 'studies.tsv', 'w')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
95 studies_table.write('\t'.join(['alias', 'status', 'accession', 'title', 'study_type',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
96 'study_abstract', 'pubmed_id', 'submission_date']) + '\n')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
97 samples_table = open(pathlib.Path(args.out_path) / 'samples.tsv', 'w')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
98 if args.viral_submission:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
99 samples_table.write('\t'.join(['alias', 'status', 'accession', 'title', 'scientific_name',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
100 'taxon_id', 'sample_description', 'collection_date',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
101 'geographic_location', 'host_common_name', 'host_subject_id',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
102 'host_health_state', 'host_sex', 'host_scientific_name',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
103 'collector_name', 'collecting_institution', 'isolate',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
104 'submission_date']) + '\n')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
105 else:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
106 samples_table.write('\t'.join(['alias', 'status', 'accession', 'title', 'scientific_name',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
107 'taxon_id', 'sample_description', 'submission_date']) + '\n')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
108
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
109 experiments_table = open(pathlib.Path(args.out_path) / 'experiments.tsv', 'w')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
110 experiments_table.write('\t'.join(['alias', 'status', 'accession', 'title', 'study_alias',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
111 'sample_alias', 'design_description', 'library_name',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
112 'library_strategy', 'library_source', 'library_selection',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
113 'library_layout', 'insert_size', 'library_construction_protocol',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
114 'platform', 'instrument_model', 'submission_date']) + '\n')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
115
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
116 runs_table = open(pathlib.Path(args.out_path) / 'runs.tsv', 'w')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
117 runs_table.write('\t'.join(['alias', 'status', 'accession', 'experiment_alias', 'file_name',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
118 'file_format', 'file_checksum', 'submission_date']) + '\n')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
119 action = args.action
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
120
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
121 # WRITE DICTIONARIES TO TABLE FILES
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
122
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
123 # ADD A TIMESTAMP TO THE ALIAS? SEEMS LIKE ENA REQUIRES ALL ENTRIES FOR A WEBIN TO HAVE UNIQUE IDS?
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
124 # dt_oobj = datetime.now(tz=None)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
125 # timestamp = dt_oobj.strftime("%Y%m%d_%H:%M:%S")
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
126 for study_alias, study in studies_dict.items():
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
127 # study_alias = study_alias + '_' + timestamp
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
128 studies_table.write('\t'.join([study_alias, action, 'ENA_accession', study['title'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
129 study['study_type'], study['study_abstract'], '',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
130 'ENA_submission_data']) + '\n') # assuming no pubmed_id
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
131 for sample_alias, sample in samples_dict.items():
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
132 # sample_alias = sample_alias + '_' + timestamp
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
133 if args.viral_submission:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
134 if sample['collector name'] == '':
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
135 sample['collector name'] = 'unknown'
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
136 samples_table.write('\t'.join([sample_alias, action, 'ena_accession', sample['title'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
137 sample['scientific_name'], 'tax_id_updated_by_ENA',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
138 sample['sample_description'], sample['collection date'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
139 sample['geographic location (country and/or sea)'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
140 sample['host common name'], 'host subject id',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
141 sample['host health state'], sample['host sex'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
142 sample['host scientific name'], sample['collector name'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
143 sample['collecting institution'], sample['isolate'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
144 'ENA_submission_date']) + '\n')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
145 else:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
146 samples_table.write('\t'.join([sample_alias, action, 'ena_accession', sample['title'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
147 sample['scientific_name'], 'tax_id_updated_by_ENA',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
148 sample['sample_description']]) + '\n')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
149 for exp_alias, exp in experiments_dict.items():
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
150 # should I check here if any experiment has a study or sample alias that is incorrect?
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
151 # (not listed in the samples or study dict)
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
152 # process the experiments for this sample
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
153 if exp['sample_alias'] == sample_alias:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
154 lib_alias = 'library_' + exp_alias + '_' + exp['sample_alias']
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
155 experiments_table.write('\t'.join([exp_alias, action, 'accession_ena', exp['title'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
156 exp['study_alias'], sample_alias,
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
157 exp['design_description'], lib_alias,
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
158 exp['library_strategy'], exp['library_source'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
159 exp['library_selection'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
160 exp['library_layout'].lower(),
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
161 str(int(exp['insert_size'])),
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
162 exp['library_construction_protocol'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
163 exp['platform'], exp['instrument_model'],
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
164 'submission_date_ENA']) + '\n')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
165 for run_alias, run in runs_dict.items():
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
166 if run['experiment_alias'] == exp_alias:
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
167 runs_table.write('\t'.join([run_alias, action, 'ena_run_accession', exp_alias,
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
168 run['file_name'], FILE_FORMAT, 'file_checksum',
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
169 'submission_date_ENA']) + '\n')
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
170 studies_table.close()
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
171 samples_table.close()
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
172 experiments_table.close()
382518f24d6d "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload commit 57b434bcf493554d060a99b65e66f274d5c00e0a"
iuc
parents:
diff changeset
173 runs_table.close()