Mercurial > repos > vandelj > giant_factor_generator
annotate src/General_functions.py @ 0:4764dc6a1019 draft
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
| author | vandelj | 
|---|---|
| date | Fri, 26 Jun 2020 09:51:15 -0400 | 
| parents | |
| children | 7a520f7169e1 | 
| rev | line source | 
|---|---|
| 0 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 1 import re | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 2 import numpy as np | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 3 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 4 def get_column_names( file_path, toNotConsider=-1, each=1): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 5 options=[] | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 6 inputfile = open(file_path) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 7 firstLine = next(inputfile).strip().split("\t") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 8 cpt=0 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 9 for i, field_component in enumerate( firstLine ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 10 if i!=toNotConsider:#to squeeze the first column | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 11 if cpt==0: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 12 options.append( ( field_component, field_component, False ) ) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 13 cpt+=1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 14 if cpt==each: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 15 cpt=0 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 16 inputfile.close() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 17 return options | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 18 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 19 def get_column_names_filteredList( file_path, toNotConsider=[], each=1): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 20 options=[] | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 21 inputfile = open(file_path) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 22 firstLine = next(inputfile).strip().split("\t") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 23 cpt=0 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 24 for i, field_component in enumerate( firstLine ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 25 if i not in toNotConsider:#to squeeze the first columns | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 26 if cpt==0: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 27 options.append( ( field_component, field_component, False ) ) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 28 cpt+=1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 29 if cpt==each: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 30 cpt=0 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 31 inputfile.close() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 32 return options | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 33 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 34 def get_column_names_mergeNumber(file_path, numberToMerge=1, toNotConsider=[]): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 35 options=[] | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 36 inputfile = open(file_path) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 37 if int(numberToMerge)>0: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 38 iHeader=0 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 39 for iCurrentLine in inputfile: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 40 iHeader=iHeader+1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 41 if iHeader>int(numberToMerge): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 42 break | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 43 currentLine=iCurrentLine.strip().split("\t") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 44 iOption=-1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 45 for i, field_component in enumerate( currentLine ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 46 if i not in toNotConsider:#to squeeze specified columns | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 47 iOption=iOption+1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 48 if iHeader==1: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 49 options.append( ( str(field_component), str(field_component), False ) ) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 50 else: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 51 options[iOption]=(options[iOption][0]+"_"+str(field_component),options[iOption][1]+"_"+str(field_component),False) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 52 else: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 53 currentLine = next(inputfile).strip().split("\t") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 54 for i, field_component in enumerate( currentLine ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 55 if i not in toNotConsider:#to squeeze specified columns | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 56 options.append( ( "Column_"+str(i), "Column_"+str(i), False ) ) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 57 inputfile.close() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 58 return options | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 59 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 60 def get_row_names( file_path, factorName ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 61 inputfile = open(file_path) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 62 firstLine = next(inputfile).strip().split("\t") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 63 iColumn=-1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 64 for i, field_component in enumerate( firstLine ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 65 if field_component==factorName:#to test | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 66 iColumn=i | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 67 options=[] | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 68 if iColumn!=-1: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 69 for nextLine in inputfile: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 70 nextLine=nextLine.strip().split("\t") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 71 if len(nextLine)>1: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 72 if (nextLine[iColumn], nextLine[iColumn], False) not in options: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 73 options.append( (nextLine[iColumn], nextLine[iColumn], False) ) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 74 inputfile.close() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 75 return options | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 76 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 77 def get_condition_file_names( file_list, toNotConsider=-1, each=1): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 78 options=[] | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 79 if not isinstance(file_list,list):#if input file is a tabular file, act as get_column_names | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 80 inputfile = open(file_list.file_name) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 81 firstLine = next(inputfile).strip().split("\t") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 82 cpt=0 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 83 for i, field_component in enumerate( firstLine ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 84 if i!=toNotConsider:#to squeeze the first column | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 85 if cpt==0: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 86 options.append( ( field_component, field_component, False ) ) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 87 cpt+=1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 88 if cpt==each: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 89 cpt=0 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 90 inputfile.close() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 91 else:#if input file is a .cel file list or a collection | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 92 if not hasattr(file_list[0],'collection'):#if it is not a collection, get name easily | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 93 for i, field_component in enumerate( file_list ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 94 options.append( ( field_component.name, field_component.name, False ) ) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 95 else:#if the file is a collection, have to get deeper in the corresponding HistoryDatasetCollectionAssociation object | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 96 for i, field_component in enumerate( file_list[0].collection.elements ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 97 options.append( ( field_component.element_identifier, field_component.element_identifier, False ) ) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 98 return options | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 99 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 100 def generateFactorFile( file_list, factor_list, outputFileName, logFile): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 101 forbidenCharacters={"*",":",",","|"} | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 102 outputfile = open(outputFileName, 'w') | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 103 outputLog = open(logFile, 'w') | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 104 sampleList=[] | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 105 if not isinstance(file_list,list): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 106 conditionNames=get_condition_file_names(file_list,0) #unique expression file, remove the first column (index=0) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 107 else : | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 108 conditionNames=get_condition_file_names(file_list) #.CEL files | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 109 for iSample, sample_component in enumerate (conditionNames): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 110 sampleList.append(str(sample_component[1])) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 111 outputLog.write("[INFO] "+str(len(sampleList))+" sample are detected as input\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 112 globalDict=dict() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 113 factorNameList=[] | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 114 firstLine="Conditions" | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 115 if len(factor_list)==0:#check if there is at least one factor available | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 116 outputLog.write("[ERROR] no factor was defined !\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 117 return 1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 118 else: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 119 for iFactor, factor_component in enumerate( factor_list ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 120 currentSampleList=list(sampleList) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 121 currentFactor=str(factor_component['factorName']) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 122 #check if factor name contains forbidden characters | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 123 for specialCharacter in forbidenCharacters: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 124 if currentFactor.find(specialCharacter)!=-1: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 125 outputLog.write("[ERROR] '"+specialCharacter+"' character is forbidden in factor name : '"+currentFactor+"'\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 126 return 4 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 127 #check if factor allready named like that | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 128 if not globalDict.get(currentFactor) is None: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 129 outputLog.write("[ERROR] '"+currentFactor+"' is used several times as factor name\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 130 return 3 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 131 globalDict[currentFactor]=dict() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 132 firstLine=firstLine+"\t"+currentFactor | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 133 factorNameList.append(currentFactor) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 134 if len(factor_component['valueList'])<=1:#check if there is at least two value available | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 135 outputLog.write("[ERROR] at least two different values are necessary for '"+currentFactor+"' factor\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 136 return 1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 137 else: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 138 for iValue, value_component in enumerate( factor_component['valueList'] ): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 139 currentValue=str(value_component['valueName']) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 140 #check if factor name contains forbidden characters | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 141 for specialCharacter in forbidenCharacters: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 142 if currentValue.find(specialCharacter)!=-1: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 143 outputLog.write("[ERROR] '"+specialCharacter+"' character is forbidden in value name : '"+currentValue+"'\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 144 return 4 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 145 currentSample=str(value_component['valueConditions']).split(",") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 146 for iSample, sample_component in enumerate (currentSample): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 147 if not sample_component in currentSampleList: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 148 outputLog.write("[ERROR] sample "+sample_component+" was assigned several times for factor '"+currentFactor+"'\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 149 return 2 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 150 currentSampleList.remove(sample_component) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 151 globalDict[currentFactor][sample_component]=currentValue | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 152 if(len(currentSampleList)>0): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 153 outputLog.write("[ERROR] for factor '"+currentFactor+"'' sample "+str(currentSampleList)+" are not assigned to any value\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 154 return 2 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 155 outputLog.write("[INFO] "+str(len(globalDict))+" factors are detected\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 156 #start writing the factor file | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 157 outputfile.write(firstLine+"\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 158 for iSample, sample_component in enumerate(sampleList): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 159 newLine=sample_component | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 160 for iFactor, factor_component in enumerate(factorNameList): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 161 newLine=newLine+"\t"+globalDict[factor_component][sample_component] | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 162 outputfile.write(newLine+"\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 163 outputfile.close() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 164 outputLog.close() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 165 return 0 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 166 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 167 def selectSubSetTable(file_path,headerLine_number,columnsToAdd,columnNamesToKeep,outputFileName,logFile): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 168 outputLog = open(logFile, 'w') | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 169 outputLog.write("[INFO] header line number : "+ headerLine_number+" lines\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 170 availableColumnsTuple=get_column_names_mergeNumber(file_path, headerLine_number) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 171 #convert tuple list as a simple array | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 172 availableColumns=[] | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 173 for iTuple, tuple_content in enumerate (availableColumnsTuple): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 174 availableColumns.append(str(tuple_content[0])) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 175 if len(availableColumns)==0: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 176 outputLog.write("[ERROR] No detected columns in input file\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 177 return 1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 178 selectedColumns=list(columnsToAdd) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 179 for iVolcano, volcano_content in enumerate(columnNamesToKeep): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 180 selectedColumns.append(availableColumns.index(volcano_content['pvalColumn'])) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 181 if volcano_content['fdrColumn'] in availableColumns: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 182 selectedColumns.append(availableColumns.index(volcano_content['fdrColumn'])) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 183 else: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 184 selectedColumns.append(0) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 185 selectedColumns.append(availableColumns.index(volcano_content['fcColumn'])) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 186 if len(selectedColumns)!=(3*len(columnNamesToKeep)+len(columnsToAdd)): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 187 outputLog.write("[ERROR] matching between input file colnames and requested column names failed\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 188 return 1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 189 outputLog.write("[INFO] columns kept : "+str(selectedColumns)+"\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 190 #start writting formatted file | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 191 inputfile = open(file_path) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 192 outputfile = open(outputFileName, 'w') | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 193 iLineCpt=-1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 194 for iCurrentLine in inputfile: | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 195 iLineCpt=iLineCpt+1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 196 if iLineCpt>=int(headerLine_number): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 197 currentLineFields=np.array(iCurrentLine.strip().split("\t")) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 198 newLine="\t".join(currentLineFields[selectedColumns]) | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 199 outputfile.write(newLine+"\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 200 if iLineCpt<int(headerLine_number): | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 201 outputLog.write("[ERROR] not enough lines in input files ("+(iLineCpt+1)+" lines)\n") | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 202 return 1 | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 203 inputfile.close() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 204 outputfile.close() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 205 outputLog.close() | 
| 
4764dc6a1019
"planemo upload for repository https://github.com/juliechevalier/GIANT/tree/master commit cb276a594444c8f32e9819fefde3a21f121d35df"
 vandelj parents: diff
changeset | 206 return 0 | 
