Mercurial > repos > pieterlukasse > prims_metabolomics
annotate query_metexp.py @ 22:cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
author | pieter.lukasse@wur.nl |
---|---|
date | Thu, 06 Mar 2014 14:29:55 +0100 |
parents | 19d8fd10248e |
children |
rev | line source |
---|---|
21
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
1 #!/usr/bin/env python |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
2 # encoding: utf-8 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
3 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
4 Module to query a set of identifications against the METabolomics EXPlorer database. |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
5 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
6 It will take the input file and for each record it will query the |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
7 molecular mass in the selected MetExp DB. If one or more compounds are found in the |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
8 MetExp DB then extra information regarding these compounds is added to the output file. |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
9 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
10 The output file is thus the input file enriched with information about |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
11 related items found in the selected MetExp DB. |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
12 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
13 import csv |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
14 import sys |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
15 import fileinput |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
16 import urllib2 |
22
cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
pieter.lukasse@wur.nl
parents:
21
diff
changeset
|
17 import time |
21
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
18 from collections import OrderedDict |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
19 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
20 __author__ = "Pieter Lukasse" |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
21 __contact__ = "pieter.lukasse@wur.nl" |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
22 __copyright__ = "Copyright, 2014, Plant Research International, WUR" |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
23 __license__ = "Apache v2" |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
24 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
25 def _process_file(in_xsv, delim='\t'): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
26 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
27 Generic method to parse a tab-separated file returning a dictionary with named columns |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
28 @param in_csv: input filename to be parsed |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
29 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
30 data = list(csv.reader(open(in_xsv, 'rU'), delimiter=delim)) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
31 return _process_data(data) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
32 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
33 def _process_data(data): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
34 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
35 header = data.pop(0) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
36 # Create dictionary with column name as key |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
37 output = OrderedDict() |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
38 for index in xrange(len(header)): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
39 output[header[index]] = [row[index] for row in data] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
40 return output |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
41 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
42 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
43 def _query_and_add_data(input_data, casid_col, formula_col, molecular_mass_col, metexp_dblink, separation_method): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
44 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
45 This method will iterate over the record in the input_data and |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
46 will enrich them with the related information found (if any) in the |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
47 MetExp Database. |
22
cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
pieter.lukasse@wur.nl
parents:
21
diff
changeset
|
48 |
cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
pieter.lukasse@wur.nl
parents:
21
diff
changeset
|
49 # TODO : could optimize this with multi-threading, see also nice example at http://stackoverflow.com/questions/2846653/python-multithreading-for-dummies |
21
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
50 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
51 merged = [] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
52 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
53 for i in xrange(len(input_data[input_data.keys()[0]])): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
54 # Get the record in same dictionary format as input_data, but containing |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
55 # a value at each column instead of a list of all values of all records: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
56 input_data_record = OrderedDict(zip(input_data.keys(), [input_data[key][i] for key in input_data.keys()])) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
57 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
58 # read the molecular mass and formula: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
59 cas_id = input_data_record[casid_col] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
60 formula = input_data_record[formula_col] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
61 molecular_mass = input_data_record[molecular_mass_col] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
62 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
63 # search for related records in MetExp: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
64 data_found = None |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
65 if cas_id != "undef": |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
66 # 1- search for other experiments where this CAS id has been found: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
67 query_link = metexp_dblink + "/find_entries/query?cas_nr="+ cas_id + "&method=" + separation_method |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
68 data_found = _fire_query_and_return_dict(query_link + "&_format_result=tsv") |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
69 data_type_found = "CAS" |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
70 if data_found == None: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
71 # 2- search for other experiments where this FORMULA has been found: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
72 query_link = metexp_dblink + "/find_entries/query?molecule_formula="+ formula + "&method=" + separation_method |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
73 data_found = _fire_query_and_return_dict(query_link + "&_format_result=tsv") |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
74 data_type_found = "FORMULA" |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
75 if data_found == None: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
76 # 3- search for other experiments where this MM has been found: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
77 query_link = metexp_dblink + "/find_entries/query?molecule_mass="+ molecular_mass + "&method=" + separation_method |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
78 data_found = _fire_query_and_return_dict(query_link + "&_format_result=tsv") |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
79 data_type_found = "MM" |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
80 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
81 if data_found == None: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
82 # If still nothing found, just add empty columns |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
83 extra_cols = ['', '','','','','','',''] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
84 else: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
85 # Add info found: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
86 extra_cols = _get_extra_info_and_link_cols(data_found, data_type_found, query_link) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
87 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
88 # Take all data and merge it into a "flat"/simple array of values: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
89 field_values_list = _merge_data(input_data_record, extra_cols) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
90 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
91 merged.append(field_values_list) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
92 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
93 # return the merged/enriched records: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
94 return merged |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
95 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
96 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
97 def _get_extra_info_and_link_cols(data_found, data_type_found, query_link): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
98 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
99 This method will go over the data found and will return a |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
100 list with the following items: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
101 - Experiment details where hits have been found : |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
102 'organism', 'tissue','experiment_name','user_name','column_type' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
103 - Link that executes same query |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
104 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
105 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
106 # set() makes a unique list: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
107 organism_set = [] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
108 tissue_set = [] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
109 experiment_name_set = [] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
110 user_name_set = [] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
111 column_type_set = [] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
112 cas_nr_set = [] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
113 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
114 if 'organism' in data_found: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
115 organism_set = set(data_found['organism']) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
116 if 'tissue' in data_found: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
117 tissue_set = set(data_found['tissue']) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
118 if 'experiment_name' in data_found: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
119 experiment_name_set = set(data_found['experiment_name']) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
120 if 'user_name' in data_found: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
121 user_name_set = set(data_found['user_name']) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
122 if 'column_type' in data_found: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
123 column_type_set = set(data_found['column_type']) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
124 if 'CAS' in data_found: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
125 cas_nr_set = set(data_found['CAS']) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
126 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
127 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
128 result = [data_type_found, |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
129 _to_xsv(organism_set), |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
130 _to_xsv(tissue_set), |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
131 _to_xsv(experiment_name_set), |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
132 _to_xsv(user_name_set), |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
133 _to_xsv(column_type_set), |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
134 _to_xsv(cas_nr_set), |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
135 #To let Excel interpret as link, use e.g. =HYPERLINK("http://stackoverflow.com", "friendly name"): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
136 "=HYPERLINK(\""+ query_link + "\", \"Link to entries found in DB \")"] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
137 return result |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
138 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
139 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
140 def _to_xsv(data_set): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
141 result = "" |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
142 for item in data_set: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
143 result = result + str(item) + "|" |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
144 return result |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
145 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
146 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
147 def _fire_query_and_return_dict(url): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
148 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
149 This method will fire the query as a web-service call and |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
150 return the results as a list of dictionary objects |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
151 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
152 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
153 try: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
154 data = urllib2.urlopen(url).read() |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
155 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
156 # transform to dictionary: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
157 result = [] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
158 data_rows = data.split("\n") |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
159 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
160 # check if there is any data in the response: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
161 if len(data_rows) <= 1 or data_rows[1].strip() == '': |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
162 # means there is only the header row...so no hits: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
163 return None |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
164 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
165 for data_row in data_rows: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
166 if not data_row.strip() == '': |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
167 row_as_list = _str_to_list(data_row, delimiter='\t') |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
168 result.append(row_as_list) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
169 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
170 # return result processed into a dict: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
171 return _process_data(result) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
172 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
173 except urllib2.HTTPError, e: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
174 raise Exception( "HTTP error for URL: " + url + " : %s - " % e.code + e.reason) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
175 except urllib2.URLError, e: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
176 raise Exception( "Network error: %s" % e.reason.args[1] + ". Administrator: please check if MetExp service [" + url + "] is accessible from your Galaxy server. ") |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
177 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
178 def _str_to_list(data_row, delimiter='\t'): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
179 result = [] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
180 for column in data_row.split(delimiter): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
181 result.append(column) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
182 return result |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
183 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
184 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
185 # alternative: ? |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
186 # s = requests.Session() |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
187 # s.verify = False |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
188 # #s.auth = (token01, token02) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
189 # resp = s.get(url, params={'name': 'anonymous'}, stream=True) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
190 # content = resp.content |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
191 # # transform to dictionary: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
192 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
193 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
194 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
195 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
196 def _merge_data(input_data_record, extra_cols): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
197 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
198 Adds the extra information to the existing data record and returns |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
199 the combined new record. |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
200 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
201 record = [] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
202 for column in input_data_record: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
203 record.append(input_data_record[column]) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
204 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
205 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
206 # add extra columns |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
207 for column in extra_cols: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
208 record.append(column) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
209 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
210 return record |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
211 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
212 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
213 def _save_data(data_rows, headers, out_csv): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
214 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
215 Writes tab-separated data to file |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
216 @param data_rows: dictionary containing merged/enriched dataset |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
217 @param out_csv: output csv file |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
218 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
219 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
220 # Open output file for writing |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
221 outfile_single_handle = open(out_csv, 'wb') |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
222 output_single_handle = csv.writer(outfile_single_handle, delimiter="\t") |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
223 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
224 # Write headers |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
225 output_single_handle.writerow(headers) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
226 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
227 # Write one line for each row |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
228 for data_row in data_rows: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
229 output_single_handle.writerow(data_row) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
230 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
231 def _get_metexp_URL(metexp_dblink_file): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
232 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
233 Read out and return the URL stored in the given file. |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
234 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
235 file_input = fileinput.input(metexp_dblink_file) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
236 try: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
237 for line in file_input: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
238 if line[0] != '#': |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
239 # just return the first line that is not a comment line: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
240 return line |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
241 finally: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
242 file_input.close() |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
243 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
244 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
245 def main(): |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
246 ''' |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
247 MetExp Query main function |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
248 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
249 The input file can be any tabular file, as long as it contains a column for the molecular mass |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
250 and one for the formula of the respective identification. These two columns are then |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
251 used to query against MetExp Database. |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
252 ''' |
22
cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
pieter.lukasse@wur.nl
parents:
21
diff
changeset
|
253 seconds_start = int(round(time.time())) |
cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
pieter.lukasse@wur.nl
parents:
21
diff
changeset
|
254 |
21
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
255 input_file = sys.argv[1] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
256 casid_col = sys.argv[2] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
257 formula_col = sys.argv[3] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
258 molecular_mass_col = sys.argv[4] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
259 metexp_dblink_file = sys.argv[5] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
260 separation_method = sys.argv[6] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
261 output_result = sys.argv[7] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
262 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
263 # Parse metexp_dblink_file to find the URL to the MetExp service: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
264 metexp_dblink = _get_metexp_URL(metexp_dblink_file) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
265 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
266 # Parse tabular input file into dictionary/array: |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
267 input_data = _process_file(input_file) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
268 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
269 # Query data against MetExp DB : |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
270 enriched_data = _query_and_add_data(input_data, casid_col, formula_col, molecular_mass_col, metexp_dblink, separation_method) |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
271 headers = input_data.keys() + ['METEXP hits for ','METEXP hits: organisms', 'METEXP hits: tissues', |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
272 'METEXP hits: experiments','METEXP hits: user names','METEXP hits: column types', 'METEXP hits: CAS nrs', 'Link to METEXP hits'] |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
273 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
274 _save_data(enriched_data, headers, output_result) |
22
cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
pieter.lukasse@wur.nl
parents:
21
diff
changeset
|
275 |
cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
pieter.lukasse@wur.nl
parents:
21
diff
changeset
|
276 seconds_end = int(round(time.time())) |
cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
pieter.lukasse@wur.nl
parents:
21
diff
changeset
|
277 print "Took " + str(seconds_end - seconds_start) + " seconds" |
cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
pieter.lukasse@wur.nl
parents:
21
diff
changeset
|
278 |
cd4f13119afa
Small fix in filters part and improvement in query_metexp time logging
pieter.lukasse@wur.nl
parents:
21
diff
changeset
|
279 |
21
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
280 |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
281 if __name__ == '__main__': |
19d8fd10248e
* Added interface to METEXP data store, including tool to fire queries in batch mode
pieter.lukasse@wur.nl
parents:
diff
changeset
|
282 main() |