|
0
|
1 #!/usr/bin/env python
|
|
|
2 # encoding: utf-8
|
|
|
3 '''
|
|
|
4 Module to combine output from the GCMS Galaxy tools RankFilter, CasLookup and MsClust
|
|
|
5 into a tabular file that can be uploaded to the MetExp database.
|
|
|
6
|
|
|
7 RankFilter, CasLookup are already combined by combine_output.py so here we will use
|
|
|
8 this result. Furthermore here the MsClust spectra file (.MSP) and one of the MsClust
|
|
|
9 quantification files are to be combined with combine_output.py result as well.
|
|
|
10
|
|
|
11 Extra calculations performed:
|
|
|
12 - The column MW is also added here and is derived from the column FORMULA found
|
|
|
13 in combine_output.py result.
|
|
|
14
|
|
|
15 So in total here we merge 3 files and calculate one new column.
|
|
|
16 '''
|
|
|
17
|
|
|
18 import csv
|
|
|
19 import sys
|
|
|
20 from collections import OrderedDict
|
|
|
21
|
|
|
22 __author__ = "Pieter Lukasse"
|
|
|
23 __contact__ = "pieter.lukasse@wur.nl"
|
|
|
24 __copyright__ = "Copyright, 2013, Plant Research International, WUR"
|
|
|
25 __license__ = "Apache v2"
|
|
|
26
|
|
|
27 def _process_data(in_csv, delim='\t'):
|
|
|
28 '''
|
|
|
29 Generic method to parse a tab-separated file returning a dictionary with named columns
|
|
|
30 @param in_csv: input filename to be parsed
|
|
|
31 '''
|
|
|
32 data = list(csv.reader(open(in_csv, 'rU'), delimiter=delim))
|
|
|
33 header = data.pop(0)
|
|
|
34 # Create dictionary with column name as key
|
|
|
35 output = OrderedDict()
|
|
|
36 for index in xrange(len(header)):
|
|
|
37 output[header[index]] = [row[index] for row in data]
|
|
|
38 return output
|
|
|
39
|
|
|
40 ONE_TO_ONE = 'one_to_one'
|
|
|
41 N_TO_ONE = 'n_to_one'
|
|
|
42
|
|
|
43 def _merge_data(set1, link_field_set1, set2, link_field_set2, compare_function, merge_function, relation_type=ONE_TO_ONE):
|
|
|
44 '''
|
|
|
45 Merges data from both input dictionaries based on the link fields. This method will
|
|
|
46 build up a new list containing the merged hits as the items.
|
|
|
47 @param set1: dictionary holding set1 in the form of N lists (one list per attribute name)
|
|
|
48 @param set2: dictionary holding set2 in the form of N lists (one list per attribute name)
|
|
|
49 '''
|
|
|
50 # TODO test for correct input files -> same link_field values should be there (test at least number of unique link_field values):
|
|
|
51 #
|
|
|
52 # if (len(set1[link_field_set1]) != len(set2[link_field_set2])):
|
|
|
53 # raise Exception('input files should have the same nr of key values ')
|
|
|
54
|
|
|
55
|
|
|
56 merged = []
|
|
|
57 processed = {}
|
|
|
58 for link_field_set1_idx in xrange(len(set1[link_field_set1])):
|
|
|
59 link_field_set1_value = set1[link_field_set1][link_field_set1_idx]
|
|
|
60 if not link_field_set1_value in processed :
|
|
|
61 # keep track of processed items to not repeat them
|
|
|
62 processed[link_field_set1_value] = link_field_set1_value
|
|
|
63
|
|
|
64 # Get the indices for current link_field_set1_value in both data-structures for proper matching
|
|
|
65 set1index = [index for index, value in enumerate(set1[link_field_set1]) if value == link_field_set1_value]
|
|
|
66 set2index = [index for index, value in enumerate(set2[link_field_set2]) if compare_function(value, link_field_set1_value)==True ]
|
|
|
67
|
|
|
68
|
|
|
69
|
|
|
70 merged_hits = []
|
|
|
71 # Combine hits
|
|
|
72 for hit in xrange(len(set1index)):
|
|
|
73 # Create records of hits to be merged ("keys" are the attribute names, so what the lines below do
|
|
|
74 # is create a new "dict" item with same "keys"/attributes, with each attribute filled with its
|
|
|
75 # corresponding value in the rankfilter or caslookup tables; i.e.
|
|
|
76 # rankfilter[key] => returns the list/array with size = nrrows, with the values for the attribute
|
|
|
77 # represented by "key". rindex[hit] => points to the row nr=hit (hit is a rownr/index)
|
|
|
78 # It just ensures the entry is made available as a plain named array for easy access.
|
|
|
79 rf_record = OrderedDict(zip(set1.keys(), [set1[key][set1index[hit]] for key in set1.keys()]))
|
|
|
80 if relation_type == ONE_TO_ONE :
|
|
|
81 cl_record = OrderedDict(zip(set2.keys(), [set2[key][set2index[hit]] for key in set2.keys()]))
|
|
|
82 else:
|
|
|
83 # is N to 1:
|
|
|
84 cl_record = OrderedDict(zip(set2.keys(), [set2[key][set2index[0]] for key in set2.keys()]))
|
|
|
85
|
|
|
86 merged_hit = merge_function(rf_record, cl_record)
|
|
|
87 merged_hits.append(merged_hit)
|
|
|
88
|
|
|
89 merged.append(merged_hits)
|
|
|
90
|
|
|
91 return merged, len(set1index)
|
|
|
92
|
|
|
93
|
|
|
94 def _compare_records(key1, key2):
|
|
|
95 '''
|
|
|
96 in this case the compare method is really simple as both keys are expected to contain
|
|
|
97 same value when records are the same
|
|
|
98 '''
|
|
|
99 if key1 == key2:
|
|
|
100 return True
|
|
|
101 else:
|
|
|
102 return False
|
|
|
103
|
|
|
104
|
|
|
105
|
|
|
106 def _merge_records(rank_caslookup_combi, msclust_quant_record):
|
|
|
107 '''
|
|
|
108 Combines single records from both the RankFilter+CasLookup combi file and from MsClust file
|
|
|
109
|
|
|
110 @param rank_caslookup_combi: rankfilter and caslookup combined record (see combine_output.py)
|
|
|
111 @param msclust_quant_record: msclust quantification + spectrum record
|
|
|
112 '''
|
|
|
113 i = 0
|
|
|
114 record = []
|
|
|
115 for column in rank_caslookup_combi:
|
|
|
116 record.append(rank_caslookup_combi[column])
|
|
|
117 i += 1
|
|
|
118
|
|
|
119 for column in msclust_quant_record:
|
|
|
120 record.append(msclust_quant_record[column])
|
|
|
121 i += 1
|
|
|
122
|
|
|
123 return record
|
|
|
124
|
|
|
125
|
|
|
126
|
|
|
127
|
|
|
128 def _save_data(data, headers, nhits, out_csv):
|
|
|
129 '''
|
|
|
130 Writes tab-separated data to file
|
|
|
131 @param data: dictionary containing merged dataset
|
|
|
132 @param out_csv: output csv file
|
|
|
133 '''
|
|
|
134
|
|
|
135 # Open output file for writing
|
|
|
136 outfile_single_handle = open(out_csv, 'wb')
|
|
|
137 output_single_handle = csv.writer(outfile_single_handle, delimiter="\t")
|
|
|
138
|
|
|
139 # Write headers
|
|
|
140 output_single_handle.writerow(headers)
|
|
|
141
|
|
|
142 # Write one line for each centrotype
|
|
|
143 for centrotype_idx in xrange(len(data)):
|
|
|
144 for hit in data[centrotype_idx]:
|
|
|
145 output_single_handle.writerow(hit)
|
|
|
146
|
|
|
147
|
|
|
148 def main():
|
|
|
149 '''
|
|
|
150 Combine Output main function
|
|
|
151
|
|
|
152 RankFilter, CasLookup are already combined by combine_output.py so here we will use
|
|
|
153 this result. Furthermore here the MsClust spectra file (.MSP) and one of the MsClust
|
|
|
154 quantification files are to be combined with combine_output.py result as well.
|
|
|
155 '''
|
|
|
156 rankfilter_and_caslookup_combined_file = sys.argv[1]
|
|
|
157 msclust_quantification_and_spectra_file = sys.argv[2]
|
|
|
158 output_csv = sys.argv[3]
|
|
|
159
|
|
|
160 # Read RankFilter and CasLookup output files
|
|
|
161 rankfilter_and_caslookup_combined = _process_data(rankfilter_and_caslookup_combined_file)
|
|
|
162 msclust_quantification_and_spectra = _process_data(msclust_quantification_and_spectra_file, ',')
|
|
|
163
|
|
|
164 merged, nhits = _merge_data(rankfilter_and_caslookup_combined, 'Centrotype',
|
|
|
165 msclust_quantification_and_spectra, 'centrotype', _compare_records, _merge_records, N_TO_ONE)
|
|
|
166 headers = rankfilter_and_caslookup_combined.keys() + msclust_quantification_and_spectra.keys()
|
|
|
167 _save_data(merged, headers, nhits, output_csv)
|
|
|
168
|
|
|
169
|
|
|
170 if __name__ == '__main__':
|
|
|
171 main()
|
