Mercurial > repos > dchristiany > data_manager_proteore

annotate data_manager/resource_building.py @ 14:f8ed6fc5f3ae draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 3787f45c66a9897be103392b34ce498ae193a803-dirty
author dchristiany
date Fri, 04 Jan 2019 06:11:49 -0500
parents a1530507fee4
children 83f57ba70416
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
1 """
10
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
2 The purpose of this script is to create source files from different databases to be used in other proteore tools
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
3 """
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
4
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
5 import os, sys, argparse, requests, time, csv, re
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
6 from io import BytesIO
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
7 from zipfile import ZipFile
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
8 from galaxy.util.json import from_json_string, to_json_string
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
9
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
10 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
11 # General functions
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
12 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
13 def unzip(url, output_file):
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
14 """
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
15 Get a zip file content from a link and unzip
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
16 """
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
17 content = requests.get(url)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
18 zipfile = ZipFile(BytesIO(content.content))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
19 output_content = ""
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
20 output_content += zipfile.open(zipfile.namelist()[0]).read()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
21 output = open(output_file, "w")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
22 output.write(output_content)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
23 output.close()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
24
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
25 def _add_data_table_entry(data_manager_dict, data_table_entry,data_table):
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
26 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {})
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
27 data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get(data_table, [])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
28 data_manager_dict['data_tables'][data_table].append(data_table_entry)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
29 return data_manager_dict
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
30
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
31 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
32 # 1. Human Protein Atlas
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
33 # - Normal tissue
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
34 # - Pathology
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
35 # - Full Atlas
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
36 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
37 def HPA_sources(data_manager_dict, tissue, target_directory):
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
38 if tissue == "HPA_normal_tissue":
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
39 tissue_name = "HPA normal tissue"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
40 url = "https://www.proteinatlas.org/download/normal_tissue.tsv.zip"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
41 elif tissue == "HPA_pathology":
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
42 tissue_name = "HPA pathology"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
43 url = "https://www.proteinatlas.org/download/pathology.tsv.zip"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
44 elif tissue == "HPA_full_atlas":
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
45 tissue_name = "HPA full atlas"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
46 url = "https://www.proteinatlas.org/download/proteinatlas.tsv.zip"
10
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
47
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
48 output_file = tissue +"_"+ time.strftime("%d-%m-%Y") + ".tsv"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
49 path = os.path.join(target_directory, output_file)
10
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
50 unzip(url, path) #download and save file
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
51 tissue_name = tissue_name + " " + time.strftime("%d/%m/%Y")
10
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
52 tissue_id = tissue_name.replace(" ","_").replace("/","-")
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
53
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
54 data_table_entry = dict(id=tissue_id, name = tissue_name, value = tissue, path = path)
12
60cb0a5ae661 planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 10
diff changeset
55 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_protein_atlas")
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
56
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
57
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
58 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
59 # 2. Peptide Atlas
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
60 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
61 def peptide_atlas_sources(data_manager_dict, tissue, target_directory):
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
62 # Define PA Human build released number (here early 2018)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
63 atlas_build_id = "472"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
64 # Define organism_id (here Human) - to be upraded when other organism added to the project
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
65 organism_id = "2"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
66 # Extract sample_category_id and output filename
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
67 sample_category_id = tissue.split("-")[0]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
68 output_file = tissue.split("-")[1] +"_"+ time.strftime("%d-%m-%Y") + ".tsv"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
69 query = "https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetPeptides?atlas_build_id=" + \
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
70 atlas_build_id + "&display_options=ShowMappings&organism_id= " + \
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
71 organism_id + "&sample_category_id=" + sample_category_id + \
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
72 "&QUERY_NAME=AT_GetPeptides&output_mode=tsv&apply_action=QUERY"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
73 download = requests.get(query)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
74 decoded_content = download.content.decode('utf-8')
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
75 cr = csv.reader(decoded_content.splitlines(), delimiter='\t')
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
76
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
77 #build dictionary by only keeping uniprot accession (not isoform) as key and sum of observations as value
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
78 uni_dict = build_dictionary(cr)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
79
10
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
80 #columns of data table peptide_atlas
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
81 date = time.strftime("%d-%m-%Y")
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
82 tissue = tissue.split("-")[1]
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
83 tissue_id = tissue+"_"+date
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
84 tissue_name = tissue_id.replace("-","/").replace("_"," ")
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
85 path = os.path.join(target_directory,output_file)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
86
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
87 with open(path,"wb") as out :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
88 w = csv.writer(out,delimiter='\t')
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
89 w.writerow(["Uniprot_AC","nb_obs"])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
90 w.writerows(uni_dict.items())
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
91
10
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
92 data_table_entry = dict(id=tissue_id, name=tissue_name, value = path, tissue = tissue)
12
60cb0a5ae661 planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 10
diff changeset
93 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_peptide_atlas")
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
94
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
95 #function to count the number of observations by uniprot id
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
96 def build_dictionary (csv) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
97 uni_dict = {}
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
98 for line in csv :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
99 if "-" not in line[2] and check_uniprot_access(line[2]) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
100 if line[2] in uni_dict :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
101 uni_dict[line[2]] += int(line[4])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
102 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
103 uni_dict[line[2]] = int(line[4])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
104
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
105 return uni_dict
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
106
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
107 #function to check if an id is an uniprot accession number : return True or False-
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
108 def check_uniprot_access (id) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
109 uniprot_pattern = re.compile("[OPQ][0-9][A-Z0-9]{3}[0-9]|[A-NR-Z][0-9]([A-Z][A-Z0-9]{2}[0-9]){1,2}")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
110 if uniprot_pattern.match(id) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
111 return True
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
112 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
113 return False
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
114
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
115
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
116
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
117 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
118 # 3. ID mapping file
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
119 #######################################################################################################
13
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
120 import ftplib, gzip, pickle
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
121 csv.field_size_limit(sys.maxsize) # to handle big files
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
122
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
123 def id_mapping_sources (data_manager_dict, species, target_directory) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
124
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
125 human = species == "human"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
126 species_dict = { "human" : "HUMAN_9606", "mouse" : "MOUSE_10090", "rat" : "RAT_10116" }
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
127 files=["idmapping_selected.tab.gz","idmapping.dat.gz"]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
128
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
129 #header
13
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
130 if human :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
131 ids_list = ["UniProt-AC","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","neXtProt","BioGrid","STRING","KEGG"]
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
132 else :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
133 ids_list = ["UniProt-AC","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","BioGrid","STRING","KEGG"]
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
134 tab = [ids_list]
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
135
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
136 #print("header ok")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
137
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
138 #selected.tab and keep only ids of interest
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
139 selected_tab_file=species_dict[species]+"_"+files[0]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
140 tab_path = download_from_uniprot_ftp(selected_tab_file,target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
141 with gzip.open(tab_path,"rt") as select :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
142 tab_reader = csv.reader(select,delimiter="\t")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
143 for line in tab_reader :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
144 tab.append([line[i] for i in [0,1,2,3,4,5,6,11,13,14,18,19,20]])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
145 os.remove(tab_path)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
146
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
147 #print("selected_tab ok")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
148
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
149 """
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
150 Supplementary ID to get from HUMAN_9606_idmapping.dat :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
151 -NextProt,BioGrid,STRING,KEGG
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
152 """
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
153
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
154 if human : ids = ['neXtProt','BioGrid','STRING','KEGG' ] #ids to get from dat_file
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
155 else : ids = ['BioGrid','STRING','KEGG' ]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
156 unidict = {}
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
157
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
158 #keep only ids of interest in dictionaries
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
159 dat_file=species_dict[species]+"_"+files[1]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
160 dat_path = download_from_uniprot_ftp(dat_file,target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
161 with gzip.open(dat_path,"rt") as dat :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
162 dat_reader = csv.reader(dat,delimiter="\t")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
163 for line in dat_reader :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
164 uniprotID=line[0] #UniProtID as key
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
165 id_type=line[1] #ID type of corresponding id, key of sub-dictionnary
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
166 cor_id=line[2] #corresponding id
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
167 if "-" not in id_type : #we don't keep isoform
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
168 if id_type in ids and uniprotID in unidict :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
169 if id_type in unidict[uniprotID] :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
170 unidict[uniprotID][id_type]= ";".join([unidict[uniprotID][id_type],cor_id]) #if there is already a value in the dictionnary
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
171 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
172 unidict[uniprotID].update({ id_type : cor_id })
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
173 elif id_type in ids :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
174 unidict[uniprotID]={id_type : cor_id}
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
175 os.remove(dat_path)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
176
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
177 #print("dat_file ok")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
178
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
179 #add ids from idmapping.dat to the final tab
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
180 for line in tab[1:] :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
181 uniprotID=line[0]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
182 if human :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
183 if uniprotID in unidict :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
184 nextprot = access_dictionary(unidict,uniprotID,'neXtProt')
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
185 if nextprot != '' : nextprot = clean_nextprot_id(nextprot,line[0])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
186 line.extend([nextprot,access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'),
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
187 access_dictionary(unidict,uniprotID,'KEGG')])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
188 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
189 line.extend(["","","",""])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
190 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
191 if uniprotID in unidict :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
192 line.extend([access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'),
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
193 access_dictionary(unidict,uniprotID,'KEGG')])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
194 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
195 line.extend(["","",""])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
196
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
197 #print ("tab ok")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
198
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
199 #add missing nextprot ID for human
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
200 if human :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
201 #build next_dict
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
202 nextprot_ids = id_list_from_nextprot_ftp("nextprot_ac_list_all.txt",target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
203 next_dict = {}
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
204 for nextid in nextprot_ids :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
205 next_dict[nextid.replace("NX_","")] = nextid
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
206 os.remove(os.path.join(target_directory,"nextprot_ac_list_all.txt"))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
207
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
208 #add missing nextprot ID
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
209 for line in tab[1:] :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
210 uniprotID=line[0]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
211 nextprotID=line[13]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
212 if nextprotID == '' and uniprotID in next_dict :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
213 line[13]=next_dict[uniprotID]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
214
13
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
215 #create empty dictionary and dictionary index
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
216 ids_dictionary, ids_dictionary_index = create_ids_dictionary(ids_list)
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
217
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
218 #fill dictionary and sub dictionaries with ids
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
219 for line in tab[1:] :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
220 for index, ids in enumerate(line) :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
221 other_id_type_index = [accession_id for accession_id in ids_dictionary_index.keys() if accession_id!=index]
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
222 for id in ids.replace(" ","").split(";") : #if there's more than one id, one key per id (example : GO)
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
223 if id not in ids_dictionary[ids_dictionary_index[index]] : #if the key is not created yet
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
224 ids_dictionary[ids_dictionary_index[index]][id]={}
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
225 for other_id_type in other_id_type_index :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
226 if ids_dictionary_index[other_id_type] not in ids_dictionary[ids_dictionary_index[index]][id] :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
227 ids_dictionary[ids_dictionary_index[index]][id][ids_dictionary_index[other_id_type]] = set(line[other_id_type].replace(" ","").split(";"))
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
228 else :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
229 ids_dictionary[ids_dictionary_index[index]][id][ids_dictionary_index[other_id_type]] |= set(line[other_id_type].replace(" ","").split(";"))
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
230 if len(ids_dictionary[ids_dictionary_index[index]][id][ids_dictionary_index[other_id_type]]) > 1 and '' in ids_dictionary[ids_dictionary_index[index]][id][ids_dictionary_index[other_id_type]] :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
231 ids_dictionary[ids_dictionary_index[index]][id][ids_dictionary_index[other_id_type]].remove('')
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
232
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
233 ##writing output files
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
234 output_file = species+"_id_mapping_"+ time.strftime("%d-%m-%Y") + ".pickle"
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
235 path = os.path.join(target_directory,output_file)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
236
13
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
237 #save ids_dictionary
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
238 with open(output_dict, 'wb') as handle:
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
239 pickle.dump(ids_dictionary, handle, protocol=pickle.HIGHEST_PROTOCOL)
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
240
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
241 name_dict={"human" : "Homo sapiens", "mouse" : "Mus musculus", "rat" : "Rattus norvegicus"}
10
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
242 name = name_dict[species]+" "+time.strftime("%d/%m/%Y")
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
243 id = species+"_id_mapping_"+ time.strftime("%d-%m-%Y")
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
244
10
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty
dchristiany
parents: 7
diff changeset
245 data_table_entry = dict(id=id, name = name, value = species, path = path)
13
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
246 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_id_mapping_dictionaries")
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
247
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
248 def download_from_uniprot_ftp(file,target_directory) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
249 ftp_dir = "pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
250 path = os.path.join(target_directory, file)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
251 ftp = ftplib.FTP("ftp.uniprot.org")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
252 ftp.login("anonymous", "anonymous")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
253 ftp.cwd(ftp_dir)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
254 ftp.retrbinary("RETR " + file, open(path, 'wb').write)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
255 ftp.quit()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
256 return (path)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
257
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
258 def id_list_from_nextprot_ftp(file,target_directory) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
259 ftp_dir = "pub/current_release/ac_lists/"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
260 path = os.path.join(target_directory, file)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
261 ftp = ftplib.FTP("ftp.nextprot.org")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
262 ftp.login("anonymous", "anonymous")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
263 ftp.cwd(ftp_dir)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
264 ftp.retrbinary("RETR " + file, open(path, 'wb').write)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
265 ftp.quit()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
266 with open(path,'r') as nextprot_ids :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
267 nextprot_ids = nextprot_ids.read().splitlines()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
268 return (nextprot_ids)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
269
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
270 #return '' if there's no value in a dictionary, avoid error
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
271 def access_dictionary (dico,key1,key2) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
272 if key1 in dico :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
273 if key2 in dico[key1] :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
274 return (dico[key1][key2])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
275 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
276 return ("")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
277 #print (key2,"not in ",dico,"[",key1,"]")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
278 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
279 return ('')
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
280
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
281 #if there are several nextprot ID for one uniprotID, return the uniprot like ID
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
282 def clean_nextprot_id (next_id,uniprotAc) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
283 if len(next_id.split(";")) > 1 :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
284 tmp = next_id.split(";")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
285 if "NX_"+uniprotAc in tmp :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
286 return ("NX_"+uniprotAc)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
287 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
288 return (tmp[1])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
289 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
290 return (next_id)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
291
13
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
292 #create empty dictionary with index for tab
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
293 def create_ids_dictionary (ids_list) :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
294 ids_dictionary = {}
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
295 for id_type in ids_list :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
296 ids_dictionary[id_type]={}
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
297 ids_dictionary_index = {}
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
298
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
299 for i,id in enumerate(ids_list) :
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
300 ids_dictionary_index[i]=id
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
301
a1530507fee4 planemo upload commit 08d8f131da0e66113519ffaa7f7e7632cb3d1eff-dirty
dchristiany
parents: 12
diff changeset
302 return(ids_dictionary,ids_dictionary_index)
0
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
303
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
304 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
305 # Main function
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
306 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
307 def main():
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
308 parser = argparse.ArgumentParser()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
309 parser.add_argument("--hpa", metavar = ("HPA_OPTION"))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
310 parser.add_argument("--peptideatlas", metavar=("SAMPLE_CATEGORY_ID"))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
311 parser.add_argument("--id_mapping", metavar = ("ID_MAPPING_SPECIES"))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
312 parser.add_argument("-o", "--output")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
313 args = parser.parse_args()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
314
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
315 data_manager_dict = {}
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
316 # Extract json file params
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
317 filename = args.output
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
318 params = from_json_string(open(filename).read())
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
319 target_directory = params[ 'output_data' ][0]['extra_files_path']
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
320 os.mkdir(target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
321
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
322 ## Download source files from HPA
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
323 try:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
324 hpa = args.hpa
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
325 except NameError:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
326 hpa = None
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
327 if hpa is not None:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
328 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
329 hpa = hpa.split(",")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
330 for hpa_tissue in hpa:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
331 HPA_sources(data_manager_dict, hpa_tissue, target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
332
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
333 ## Download source file from Peptide Atlas query
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
334 try:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
335 peptide_atlas = args.peptideatlas
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
336 except NameError:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
337 peptide_atlas = None
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
338 if peptide_atlas is not None:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
339 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
340 peptide_atlas = peptide_atlas.split(",")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
341 for pa_tissue in peptide_atlas:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
342 peptide_atlas_sources(data_manager_dict, pa_tissue, target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
343
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
344 ## Download ID_mapping source file from Uniprot
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
345 try:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
346 id_mapping=args.id_mapping
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
347 except NameError:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
348 id_mapping = None
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
349 if id_mapping is not None:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
350 id_mapping = id_mapping .split(",")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
351 for species in id_mapping :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
352 id_mapping_sources(data_manager_dict, species, target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
353
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
354 #save info to json file
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
355 filename = args.output
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
356 open(filename, 'wb').write(to_json_string(data_manager_dict))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
357
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
358 if __name__ == "__main__":
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty
dchristiany
parents:
diff changeset
359 main()