Mercurial > repos > pieterlukasse > prims_metabolomics2

annotate match_library.py @ 25:9f03c8587d6b draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
MetExp msclust upload format changed from tabular to csv
author linda-bakker
date Fri, 24 Aug 2018 09:56:05 -0400
parents 43902da5d00e
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
21
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
1 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
2 Containing functions are called from Galaxy to populate lists/checkboxes with selectable items
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
3 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
4 import csv
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
5 import glob
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
6 import os
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
7
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
8
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
9 __author__ = "Marcel Kempenaar"
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
10 __contact__ = "brs@nbic.nl"
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
11 __copyright__ = "Copyright, 2012, Netherlands Bioinformatics Centre"
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
12 __license__ = "MIT"
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
13
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
14 def get_column_type(library_file):
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
15 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
16 Returns a Galaxy formatted list of tuples containing all possibilities for the
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
17 GC-column types. Used by the library_lookup.xml tool
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
18 @param library_file: given library file from which the list of GC-column types is extracted
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
19 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
20 if library_file == "":
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
21 galaxy_output = [("", "", False)]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
22 else:
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
23 (data, header) = read_library(library_file)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
24
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
25 if 'columntype' not in header:
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
26 raise IOError('(get_column_type) Missing columns in ', library_file)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
27
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
28 # Filter data on column type
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
29 column_type = header.index("columntype")
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
30 amounts_in_list_dict = count_occurrence([row[column_type] for row in data])
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
31 galaxy_output = [(str(a) + "(" + str(b) + ")", a, False) for a, b in amounts_in_list_dict.items()]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
32
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
33 return(galaxy_output)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
34
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
35
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
36 def filter_column(library_file, column_type_name):
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
37 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
38 Filters the Retention Index database on column type
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
39 @param library_file: file containing the database
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
40 @param column_type_name: column type to filter on
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
41 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
42 if library_file == "":
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
43 galaxy_output = [("", "", False)]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
44 else:
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
45 (data, header) = read_library(library_file)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
46
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
47 if ('columntype' not in header or
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
48 'columnphasetype' not in header):
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
49 raise IOError('(filter_column) Missing columns in ', library_file)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
50
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
51 column_type = header.index("columntype")
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
52 statphase = header.index("columnphasetype")
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
53
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
54 # Filter data on colunn type name
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
55 statphase_list = [line[statphase] for line in data if line[column_type] == column_type_name]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
56 amounts_in_list_dict = count_occurrence(statphase_list)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
57 galaxy_output = [(str(a) + "(" + str(b) + ")", a, False)for a, b in amounts_in_list_dict.items()]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
58
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
59 return(sorted(galaxy_output))
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
60
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
61
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
62 def filter_column2(library_file, column_type_name, statphase):
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
63 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
64 Filters the Retention Index database on column type
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
65 @param library_file: file containing the database
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
66 @param column_type_name: column type to filter on
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
67 @param statphase: stationary phase of the column to filter on
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
68 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
69 if library_file == "":
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
70 galaxy_output = [("", "", False)]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
71 else:
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
72 (data, header) = read_library(library_file)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
73
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
74 if ('columntype' not in header or
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
75 'columnphasetype' not in header or
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
76 'columnname' not in header):
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
77 raise IOError('(filter_column2) Missing columns in ', library_file)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
78
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
79 column_type_column = header.index("columntype")
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
80 statphase_column = header.index("columnphasetype")
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
81 column_name_column = header.index("columnname")
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
82
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
83 # Filter data on given column type name and stationary phase
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
84 statphase_list = [line[column_name_column] for line in data if line[column_type_column] == column_type_name and
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
85 line[statphase_column] == statphase]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
86 amounts_in_list_dict = count_occurrence(statphase_list)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
87 galaxy_output = [(str(a) + "(" + str(b) + ")", a, False)for a, b in amounts_in_list_dict.items()]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
88
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
89 return(sorted(galaxy_output))
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
90
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
91
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
92 def read_library(filename):
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
93 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
94 Reads a CSV file and returns its contents and a normalized header
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
95 @param filename: file to read
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
96 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
97 data = list(csv.reader(open(filename, 'rU'), delimiter='\t'))
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
98 header_clean = [i.lower().strip().replace(".", "").replace("%", "") for i in data.pop(0)]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
99 return(data, header_clean)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
100
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
101
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
102
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
103 def get_directory_files(dir_name):
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
104 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
105 Reads the directory and
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
106 returns the list of .txt files found as a dictionary
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
107 with file name and full path so that it can
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
108 fill a Galaxy drop-down combo box.
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
109
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
110 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
111 files = glob.glob(dir_name + "/*.*")
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
112 if len(files) == 0:
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
113 # Configuration error: no library files found in <galaxy-home-dir>/" + dir_name :
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
114 galaxy_output = [("Configuration error: expected file not found in <galaxy-home-dir>/" + dir_name, "", False)]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
115 else:
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
116 galaxy_output = [(str(get_file_name_no_ext(file_name)), str(os.path.abspath(file_name)), False) for file_name in files]
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
117 return(galaxy_output)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
118
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
119 def get_file_name_no_ext(full_name):
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
120 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
121 returns just the last part of the name
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
122 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
123 simple_name = os.path.basename(full_name)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
124 base, ext = os.path.splitext(simple_name)
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
125 return base
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
126
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
127
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
128 def count_occurrence(data_list):
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
129 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
130 Counts occurrences in a list and returns a dict with item:occurrence
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
131 @param data_list: list to count items from
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
132 '''
43902da5d00e changed match_library location again
linda.bakker@wur.nl <linda.bakker@wur.nl>
parents:
diff changeset
133 return dict((key, data_list.count(key)) for key in set(data_list))