Mercurial > repos > vandelj > giant_gsea_format
annotate src/General_functions.py @ 2:738ed1deed23 draft default tip
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit 89b7d0e8812f53222691cffe29bd48be6519829d"
author | vandelj |
---|---|
date | Fri, 25 Sep 2020 08:51:31 +0000 |
parents | d72f1bc5ce9e |
children |
rev | line source |
---|---|
0
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
1 import re |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
2 import numpy as np |
1
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
3 import galaxy.model |
0
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
4 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
5 def get_column_names( file_path, toNotConsider=-1, each=1): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
6 options=[] |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
7 inputfile = open(file_path) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
8 firstLine = next(inputfile).strip().split("\t") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
9 cpt=0 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
10 for i, field_component in enumerate( firstLine ): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
11 if i!=toNotConsider:#to squeeze the first column |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
12 if cpt==0: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
13 options.append( ( field_component, field_component, False ) ) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
14 cpt+=1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
15 if cpt==each: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
16 cpt=0 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
17 inputfile.close() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
18 return options |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
19 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
20 def get_column_names_filteredList( file_path, toNotConsider=[], each=1): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
21 options=[] |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
22 inputfile = open(file_path) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
23 firstLine = next(inputfile).strip().split("\t") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
24 cpt=0 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
25 for i, field_component in enumerate( firstLine ): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
26 if i not in toNotConsider:#to squeeze the first columns |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
27 if cpt==0: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
28 options.append( ( field_component, field_component, False ) ) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
29 cpt+=1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
30 if cpt==each: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
31 cpt=0 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
32 inputfile.close() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
33 return options |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
34 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
35 def get_column_names_mergeNumber(file_path, numberToMerge=1, toNotConsider=[]): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
36 options=[] |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
37 inputfile = open(file_path) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
38 if int(numberToMerge)>0: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
39 iHeader=0 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
40 for iCurrentLine in inputfile: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
41 iHeader=iHeader+1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
42 if iHeader>int(numberToMerge): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
43 break |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
44 currentLine=iCurrentLine.strip().split("\t") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
45 iOption=-1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
46 for i, field_component in enumerate( currentLine ): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
47 if i not in toNotConsider:#to squeeze specified columns |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
48 iOption=iOption+1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
49 if iHeader==1: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
50 options.append( ( str(field_component), str(field_component), False ) ) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
51 else: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
52 options[iOption]=(options[iOption][0]+"_"+str(field_component),options[iOption][1]+"_"+str(field_component),False) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
53 else: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
54 currentLine = next(inputfile).strip().split("\t") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
55 for i, field_component in enumerate( currentLine ): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
56 if i not in toNotConsider:#to squeeze specified columns |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
57 options.append( ( "Column_"+str(i), "Column_"+str(i), False ) ) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
58 inputfile.close() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
59 return options |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
60 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
61 def get_row_names( file_path, factorName ): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
62 inputfile = open(file_path) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
63 firstLine = next(inputfile).strip().split("\t") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
64 iColumn=-1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
65 for i, field_component in enumerate( firstLine ): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
66 if field_component==factorName:#to test |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
67 iColumn=i |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
68 options=[] |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
69 if iColumn!=-1: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
70 for nextLine in inputfile: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
71 nextLine=nextLine.strip().split("\t") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
72 if len(nextLine)>1: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
73 if (nextLine[iColumn], nextLine[iColumn], False) not in options: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
74 options.append( (nextLine[iColumn], nextLine[iColumn], False) ) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
75 inputfile.close() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
76 return options |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
77 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
78 def get_condition_file_names( file_list, toNotConsider=-1, each=1): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
79 options=[] |
1
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
80 if not (isinstance(file_list,list) or isinstance(file_list,galaxy.model.HistoryDatasetCollectionAssociation) or isinstance(file_list,galaxy.model.DatasetCollection)) :#if input file is a tabular file, act as get_column_names |
0
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
81 inputfile = open(file_list.file_name) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
82 firstLine = next(inputfile).strip().split("\t") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
83 cpt=0 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
84 for i, field_component in enumerate( firstLine ): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
85 if i!=toNotConsider:#to squeeze the first column |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
86 if cpt==0: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
87 options.append( ( field_component, field_component, False ) ) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
88 cpt+=1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
89 if cpt==each: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
90 cpt=0 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
91 inputfile.close() |
1
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
92 else:#if input file is a .cel file list, a DatasetCollection or a HistoryDatasetCollectionAssociation |
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
93 if isinstance(file_list,list):#if it is a list, retrieve names easily |
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
94 for i, field_component in enumerate( file_list ): |
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
95 options.append( ( field_component.name, field_component.name, False ) ) |
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
96 else:#if the file is a DatasetCollection, have to get deeper in the corresponding DatasetCollection object |
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
97 if isinstance(file_list,galaxy.model.DatasetCollection):#if it is a list, retrieve names easily |
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
98 for i, field_component in enumerate( file_list.elements ): |
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
99 options.append( ( field_component.element_identifier, field_component.element_identifier, False ) ) |
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
100 else:#if the file is a HistoryDatasetCollectionAssociation, have to get a little bit deeper in the corresponding HistoryDatasetCollectionAssociation object |
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
101 for i, field_component in enumerate( file_list.collection.elements ): |
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
102 options.append( ( field_component.element_identifier, field_component.element_identifier, False ) ) |
0
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
103 return options |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
104 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
105 def generateFactorFile( file_list, factor_list, outputFileName, logFile): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
106 forbidenCharacters={"*",":",",","|"} |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
107 outputfile = open(outputFileName, 'w') |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
108 outputLog = open(logFile, 'w') |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
109 sampleList=[] |
1
d72f1bc5ce9e
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit e2b27d6ff2eab66454f984dbf1a519192f41db97"
vandelj
parents:
0
diff
changeset
|
110 conditionNames=get_condition_file_names(file_list,0) #if it's a unique expression file, remove the first column (index=0) |
0
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
111 for iSample, sample_component in enumerate (conditionNames): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
112 sampleList.append(str(sample_component[1])) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
113 outputLog.write("[INFO] "+str(len(sampleList))+" sample are detected as input\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
114 globalDict=dict() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
115 factorNameList=[] |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
116 firstLine="Conditions" |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
117 if len(factor_list)==0:#check if there is at least one factor available |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
118 outputLog.write("[ERROR] no factor was defined !\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
119 return 1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
120 else: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
121 for iFactor, factor_component in enumerate( factor_list ): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
122 currentSampleList=list(sampleList) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
123 currentFactor=str(factor_component['factorName']) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
124 #check if factor name contains forbidden characters |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
125 for specialCharacter in forbidenCharacters: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
126 if currentFactor.find(specialCharacter)!=-1: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
127 outputLog.write("[ERROR] '"+specialCharacter+"' character is forbidden in factor name : '"+currentFactor+"'\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
128 return 4 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
129 #check if factor allready named like that |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
130 if not globalDict.get(currentFactor) is None: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
131 outputLog.write("[ERROR] '"+currentFactor+"' is used several times as factor name\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
132 return 3 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
133 globalDict[currentFactor]=dict() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
134 firstLine=firstLine+"\t"+currentFactor |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
135 factorNameList.append(currentFactor) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
136 if len(factor_component['valueList'])<=1:#check if there is at least two value available |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
137 outputLog.write("[ERROR] at least two different values are necessary for '"+currentFactor+"' factor\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
138 return 1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
139 else: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
140 for iValue, value_component in enumerate( factor_component['valueList'] ): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
141 currentValue=str(value_component['valueName']) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
142 #check if factor name contains forbidden characters |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
143 for specialCharacter in forbidenCharacters: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
144 if currentValue.find(specialCharacter)!=-1: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
145 outputLog.write("[ERROR] '"+specialCharacter+"' character is forbidden in value name : '"+currentValue+"'\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
146 return 4 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
147 currentSample=str(value_component['valueConditions']).split(",") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
148 for iSample, sample_component in enumerate (currentSample): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
149 if not sample_component in currentSampleList: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
150 outputLog.write("[ERROR] sample "+sample_component+" was assigned several times for factor '"+currentFactor+"'\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
151 return 2 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
152 currentSampleList.remove(sample_component) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
153 globalDict[currentFactor][sample_component]=currentValue |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
154 if(len(currentSampleList)>0): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
155 outputLog.write("[ERROR] for factor '"+currentFactor+"'' sample "+str(currentSampleList)+" are not assigned to any value\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
156 return 2 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
157 outputLog.write("[INFO] "+str(len(globalDict))+" factors are detected\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
158 #start writing the factor file |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
159 outputfile.write(firstLine+"\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
160 for iSample, sample_component in enumerate(sampleList): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
161 newLine=sample_component |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
162 for iFactor, factor_component in enumerate(factorNameList): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
163 newLine=newLine+"\t"+globalDict[factor_component][sample_component] |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
164 outputfile.write(newLine+"\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
165 outputfile.close() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
166 outputLog.close() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
167 return 0 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
168 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
169 def selectSubSetTable(file_path,headerLine_number,columnsToAdd,columnNamesToKeep,outputFileName,logFile): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
170 outputLog = open(logFile, 'w') |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
171 outputLog.write("[INFO] header line number : "+ headerLine_number+" lines\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
172 availableColumnsTuple=get_column_names_mergeNumber(file_path, headerLine_number) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
173 #convert tuple list as a simple array |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
174 availableColumns=[] |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
175 for iTuple, tuple_content in enumerate (availableColumnsTuple): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
176 availableColumns.append(str(tuple_content[0])) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
177 if len(availableColumns)==0: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
178 outputLog.write("[ERROR] No detected columns in input file\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
179 return 1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
180 selectedColumns=list(columnsToAdd) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
181 for iVolcano, volcano_content in enumerate(columnNamesToKeep): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
182 selectedColumns.append(availableColumns.index(volcano_content['pvalColumn'])) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
183 if volcano_content['fdrColumn'] in availableColumns: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
184 selectedColumns.append(availableColumns.index(volcano_content['fdrColumn'])) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
185 else: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
186 selectedColumns.append(0) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
187 selectedColumns.append(availableColumns.index(volcano_content['fcColumn'])) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
188 if len(selectedColumns)!=(3*len(columnNamesToKeep)+len(columnsToAdd)): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
189 outputLog.write("[ERROR] matching between input file colnames and requested column names failed\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
190 return 1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
191 outputLog.write("[INFO] columns kept : "+str(selectedColumns)+"\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
192 #start writting formatted file |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
193 inputfile = open(file_path) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
194 outputfile = open(outputFileName, 'w') |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
195 iLineCpt=-1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
196 for iCurrentLine in inputfile: |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
197 iLineCpt=iLineCpt+1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
198 if iLineCpt>=int(headerLine_number): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
199 currentLineFields=np.array(iCurrentLine.strip().split("\t")) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
200 newLine="\t".join(currentLineFields[selectedColumns]) |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
201 outputfile.write(newLine+"\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
202 if iLineCpt<int(headerLine_number): |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
203 outputLog.write("[ERROR] not enough lines in input files ("+(iLineCpt+1)+" lines)\n") |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
204 return 1 |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
205 inputfile.close() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
206 outputfile.close() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
207 outputLog.close() |
3022feec50fe
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
vandelj
parents:
diff
changeset
|
208 return 0 |