Mercurial > repos > proteore > proteore_data_manager

annotate data_manager/resource_building.py @ 4:e967a99d66b3 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
author proteore
date Thu, 23 Jan 2020 08:38:02 -0500
parents af0250fd023c
children b05fa99ddda2
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
1 # -*- coding: utf-8 -*-
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
2 """
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
3 The purpose of this script is to create source files from different databases to be used in other proteore tools
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
4 """
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
5
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
6 import os, shutil, sys, argparse, requests, time, csv, re, json, shutil, zipfile, subprocess
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
7 from io import BytesIO
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
8 from zipfile import ZipFile
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
9 from galaxy.util.json import from_json_string, to_json_string
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
10
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
11 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
12 # General functions
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
13 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
14 def unzip(url, output_file):
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
15 """
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
16 Get a zip file content from a link and unzip
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
17 """
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
18 content = requests.get(url)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
19 zipfile = ZipFile(BytesIO(content.content))
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
20 output_content = ""
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
21 output_content += zipfile.open(zipfile.namelist()[0]).read()
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
22 output = open(output_file, "w")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
23 output.write(output_content)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
24 output.close()
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
25
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
26 def _add_data_table_entry(data_manager_dict, data_table_entry,data_table):
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
27 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {})
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
28 data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get(data_table, [])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
29 data_manager_dict['data_tables'][data_table].append(data_table_entry)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
30 return data_manager_dict
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
31
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
32 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
33 # 1. Human Protein Atlas
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
34 # - Normal tissue
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
35 # - Pathology
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
36 # - Full Atlas
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
37 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
38 def HPA_sources(data_manager_dict, tissue, target_directory):
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
39 if tissue == "HPA_normal_tissue":
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
40 tissue_name = "HPA normal tissue"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
41 url = "https://www.proteinatlas.org/download/normal_tissue.tsv.zip"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
42 table = "proteore_protein_atlas_normal_tissue"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
43 elif tissue == "HPA_pathology":
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
44 tissue_name = "HPA pathology"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
45 url = "https://www.proteinatlas.org/download/pathology.tsv.zip"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
46 table = "proteore_protein_atlas_tumor_tissue"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
47 elif tissue == "HPA_full_atlas":
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
48 tissue_name = "HPA full atlas"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
49 url = "https://www.proteinatlas.org/download/proteinatlas.tsv.zip"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
50 table = "proteore_protein_full_atlas"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
51
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
52 output_file = tissue +"_"+ time.strftime("%d-%m-%Y") + ".tsv"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
53 path = os.path.join(target_directory, output_file)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
54 unzip(url, path) #download and save file
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
55 tissue_name = tissue_name + " " + time.strftime("%d/%m/%Y")
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
56 release = tissue_name.replace(" ","_").replace("/","-")
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
57 id = str(10000000000 - int(time.strftime("%Y%m%d")))
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
58
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
59
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
60 data_table_entry = dict(id=id, release=release, name = tissue_name, tissue = tissue, value = path)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
61 _add_data_table_entry(data_manager_dict, data_table_entry, table)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
62
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
63
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
64 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
65 # 2. Peptide Atlas
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
66 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
67 def peptide_atlas_sources(data_manager_dict, tissue, date, target_directory):
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
68 # Define organism_id (here Human) - to be upraded when other organism added to the project
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
69 organism_id = "2"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
70 # Extract sample_category_id and output filename
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
71 tissue=tissue.split(".")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
72 sample_category_id = tissue[0]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
73 tissue_name = tissue[1]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
74 output_file = tissue_name+"_"+date + ".tsv"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
75
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
76 query="https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id="+ \
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
77 sample_category_id+"&display_options=ShowAbundances&organism_id="+organism_id+ \
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
78 "&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf\
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
79 &QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
80
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
81 with requests.Session() as s:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
82 download = s.get(query)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
83 decoded_content = download.content.decode('utf-8')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
84 cr = csv.reader(decoded_content.splitlines(), delimiter='\t')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
85
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
86 uni_dict = build_dictionary(cr)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
87
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
88 #columns of data table peptide_atlas
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
89 tissue_id = tissue_name+"_"+date
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
90 name = tissue_id.replace("-","/").replace("_"," ")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
91 path = os.path.join(target_directory,output_file)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
92
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
93 with open(path,"w") as out :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
94 w = csv.writer(out,delimiter='\t')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
95 w.writerow(["Uniprot_AC","nb_obs"])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
96 w.writerows(uni_dict.items())
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
97
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
98 data_table_entry = dict(id=tissue_id, name=name, value = path, tissue = tissue_name)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
99 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_peptide_atlas")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
100
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
101 #function to count the number of observations by uniprot id
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
102 def build_dictionary (csv) :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
103 uni_dict = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
104 for line in csv :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
105 if "-" not in line[0] and check_uniprot_access(line[0]) :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
106 if line[0] in uni_dict :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
107 uni_dict[line[0]] += int(line[5])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
108 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
109 uni_dict[line[0]] = int(line[5])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
110
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
111 return uni_dict
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
112
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
113 #function to check if an id is an uniprot accession number : return True or False-
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
114 def check_uniprot_access (id) :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
115 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}")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
116 if uniprot_pattern.match(id) :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
117 return True
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
118 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
119 return False
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
120
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
121 def check_entrez_geneid (id) :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
122 entrez_pattern = re.compile("[0-9]+|[A-Z]{1,2}_[0-9]+|[A-Z]{1,2}_[A-Z]{1,4}[0-9]+")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
123 if entrez_pattern.match(id) :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
124 return True
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
125 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
126 return False
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
127
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
128 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
129 # 3. ID mapping file
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
130 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
131 import ftplib, gzip
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
132 csv.field_size_limit(sys.maxsize) # to handle big files
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
133
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
134 def id_mapping_sources (data_manager_dict, species, target_directory, tool_data_path) :
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
135
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
136 human = species == "Human"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
137 species_dict = { "Human" : "HUMAN_9606", "Mouse" : "MOUSE_10090", "Rat" : "RAT_10116" }
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
138 files=["idmapping_selected.tab.gz","idmapping.dat.gz"]
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
139 archive = os.path.join(tool_data_path, "id_mapping/ID_mapping_archive_"+species+"_"+str(time.strftime("%Y%m%d")))
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
140 if os.path.isdir(archive) is False : os.mkdir(archive)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
141
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
142 #header
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
143 if human : tab = [["UniProt-AC","UniProt-AC_reviewed","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","neXtProt","BioGrid","STRING","KEGG",'Gene_Name']]
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
144 else : tab = [["UniProt-AC","UniProt-AC_reviewed","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","BioGrid","STRING","KEGG",'Gene_Name']]
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
145
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
146 #get selected.tab and keep only ids of interest
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
147 selected_tab_file=species_dict[species]+"_"+files[0]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
148 tab_path = download_from_uniprot_ftp(selected_tab_file,target_directory)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
149 with gzip.open(tab_path,"rt") as select :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
150 tab_reader = csv.reader(select,delimiter="\t")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
151 for line in tab_reader :
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
152 tab.append([line[0]]+[line[i] for i in [0,1,2,3,4,5,6,11,13,14,18,19,20]])
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
153 if os.path.exists(os.path.join(archive,tab_path.split("/")[-1])) : os.remove(os.path.join(archive,tab_path.split("/")[-1]))
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
154 shutil.move(tab_path, archive)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
155 #print("selected_tab ok")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
156
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
157 #get uniprot-AC reviewed
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
158 organism = species_dict[species].split("_")[1]
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
159 query = "https://www.uniprot.org/uniprot/?query=reviewed:yes+AND+organism:"+organism+"&format=list"
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
160
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
161 with requests.Session() as s:
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
162 download = s.get(query)
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
163 decoded_content = download.content.decode('utf-8')
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
164 uniprot_reviewed_list = decoded_content.splitlines()
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
165
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
166 #save reviewed list
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
167 reviewed_list_path = os.path.join(archive,'uniprot_reviewed_list.txt')
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
168 with open(reviewed_list_path,'w') as reviewed_list_file:
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
169 for id in uniprot_reviewed_list:
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
170 reviewed_list_file.write(id+"\n")
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
171
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
172 #remove unreviewed uniprot-AC
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
173 for line in tab[1:]:
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
174 UniProtAC = line[1]
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
175 if UniProtAC not in uniprot_reviewed_list :
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
176 line[1]=""
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
177
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
178 """
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
179 Supplementary ID to get from HUMAN_9606_idmapping.dat :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
180 -NextProt,BioGrid,STRING,KEGG
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
181 """
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
182
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
183 #there's more id type for human
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
184 if human : ids = ['neXtProt','BioGrid','STRING','KEGG','Gene_Name' ] #ids to get from dat_file
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
185 else : ids = ['BioGrid','STRING','KEGG','Gene_Name' ]
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
186 unidict = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
187
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
188 #keep only ids of interest in dictionaries
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
189 dat_file = species_dict[species]+"_"+files[1]
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
190 dat_path = download_from_uniprot_ftp(dat_file,target_directory)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
191 with gzip.open(dat_path,"rt") as dat :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
192 dat_reader = csv.reader(dat,delimiter="\t")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
193 for line in dat_reader :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
194 uniprotID=line[0] #UniProtID as key
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
195 id_type=line[1] #ID type of corresponding id, key of sub-dictionnary
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
196 cor_id=line[2] #corresponding id
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
197 if "-" not in id_type : #we don't keep isoform
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
198 if id_type in ids and uniprotID in unidict :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
199 if id_type in unidict[uniprotID] :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
200 unidict[uniprotID][id_type]= ";".join([unidict[uniprotID][id_type],cor_id]) #if there is already a value in the dictionnary
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
201 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
202 unidict[uniprotID].update({ id_type : cor_id })
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
203 elif id_type in ids :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
204 unidict[uniprotID]={id_type : cor_id}
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
205 if os.path.exists(os.path.join(archive,dat_path.split("/")[-1])) : os.remove(os.path.join(archive,dat_path.split("/")[-1]))
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
206 shutil.move(dat_path, archive)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
207
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
208 #print("dat_file ok")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
209
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
210 #add ids from idmapping.dat to the final tab
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
211 for line in tab[1:] :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
212 uniprotID=line[0]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
213 if human :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
214 if uniprotID in unidict :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
215 nextprot = access_dictionary(unidict,uniprotID,'neXtProt')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
216 if nextprot != '' : nextprot = clean_nextprot_id(nextprot,line[0])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
217 line.extend([nextprot,access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'),
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
218 access_dictionary(unidict,uniprotID,'KEGG'),access_dictionary(unidict,uniprotID,'Gene_Name')])
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
219 else :
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
220 line.extend(["","","","",""])
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
221 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
222 if uniprotID in unidict :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
223 line.extend([access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'),
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
224 access_dictionary(unidict,uniprotID,'KEGG'),access_dictionary(unidict,uniprotID,'Gene_Name')])
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
225 else :
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
226 line.extend(["","","",""])
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
227
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
228 #print ("tab ok")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
229
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
230 #add missing nextprot ID for human or replace old ones
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
231 if human :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
232 #build next_dict
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
233 nextprot_path = id_list_from_nextprot_ftp("nextprot_ac_list_all.txt",target_directory)
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
234 with open(nextprot_path,'r') as nextprot_ids :
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
235 nextprot_ids = nextprot_ids.read().splitlines()
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
236 if os.path.exists(os.path.join(archive,nextprot_path.split("/")[-1])) : os.remove(os.path.join(archive,nextprot_path.split("/")[-1]))
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
237 shutil.move(nextprot_path,archive)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
238 next_dict = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
239 for nextid in nextprot_ids :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
240 next_dict[nextid.replace("NX_","")] = nextid
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
241 #os.remove(os.path.join(target_directory,"nextprot_ac_list_all.txt"))
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
242
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
243 #add missing nextprot ID
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
244 for line in tab[1:] :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
245 uniprotID=line[0]
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
246 nextprotID=line[14]
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
247 if uniprotID in next_dict and (nextprotID == '' or (nextprotID != "NX_"+uniprotID and next_dict[uniprotID] == "NX_"+uniprotID)) :
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
248 line[14]=next_dict[uniprotID]
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
249
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
250 output_file = species+"_id_mapping_"+ time.strftime("%d-%m-%Y") + ".tsv"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
251 path = os.path.join(target_directory,output_file)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
252
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
253 with open(path,"w") as out :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
254 w = csv.writer(out,delimiter='\t')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
255 w.writerows(tab)
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
256
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
257 subprocess.call(['tar', '-czvf', archive+".tar.gz", archive])
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
258 shutil.rmtree(archive, ignore_errors=True)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
259
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
260 name_dict={"Human" : "Homo sapiens", "Mouse" : "Mus musculus", "Rat" : "Rattus norvegicus"}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
261 name = species +" (" + name_dict[species]+" "+time.strftime("%d/%m/%Y")+")"
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
262 release = species+"_id_mapping_"+ time.strftime("%d-%m-%Y")
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
263 id = str(10000000000 - int(time.strftime("%Y%m%d"))) #new ids must be inferior to previous id -> sort by <filter> in xml only in descending order
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
264
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
265 data_table_entry = dict(id=id, release=release , name = name, species = species, value = path)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
266 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_id_mapping_"+species)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
267
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
268 def download_from_uniprot_ftp(file,target_directory) :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
269 ftp_dir = "pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
270 path = os.path.join(target_directory, file)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
271 ftp = ftplib.FTP("ftp.uniprot.org")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
272 ftp.login("anonymous", "anonymous")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
273 ftp.cwd(ftp_dir)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
274 ftp.retrbinary("RETR " + file, open(path, 'wb').write)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
275 ftp.quit()
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
276 return (path)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
277
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
278 def id_list_from_nextprot_ftp(file,target_directory) :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
279 ftp_dir = "pub/current_release/ac_lists/"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
280 path = os.path.join(target_directory, file)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
281 ftp = ftplib.FTP("ftp.nextprot.org")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
282 ftp.login("anonymous", "anonymous")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
283 ftp.cwd(ftp_dir)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
284 ftp.retrbinary("RETR " + file, open(path, 'wb').write)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
285 ftp.quit()
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
286
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
287 return (path)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
288
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
289 #return '' if there's no value in a dictionary, avoid error
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
290 def access_dictionary (dico,key1,key2) :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
291 if key1 in dico :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
292 if key2 in dico[key1] :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
293 return (dico[key1][key2])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
294 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
295 return ("")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
296 #print (key2,"not in ",dico,"[",key1,"]")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
297 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
298 return ('')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
299
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
300 #if there are several nextprot ID for one uniprotID, return the uniprot like ID
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
301 def clean_nextprot_id (next_id,uniprotAc) :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
302 if len(next_id.split(";")) > 1 :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
303 tmp = next_id.split(";")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
304 if "NX_"+uniprotAc in tmp :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
305 return ("NX_"+uniprotAc)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
306 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
307 return (tmp[1])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
308 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
309 return (next_id)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
310
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
311
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
312 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
313 # 4. Build protein interaction maps files
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
314 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
315
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
316 def get_interactant_name(line,dico):
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
317
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
318 if line[0] in dico :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
319 interactant_A = dico[line[0]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
320 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
321 interactant_A = "NA"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
322
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
323 if line[1] in dico :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
324 interactant_B = dico[line[1]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
325 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
326 interactant_B = "NA"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
327
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
328 return interactant_A, interactant_B
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
329
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
330 def PPI_ref_files(data_manager_dict, species, interactome, target_directory):
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
331
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
332 species_dict={'Human':'Homo sapiens',"Mouse":"Mus musculus","Rat":"Rattus norvegicus"}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
333
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
334 ##BioGRID
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
335 if interactome=="biogrid":
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
336
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
337 tab2_link="https://downloads.thebiogrid.org/Download/BioGRID/Release-Archive/BIOGRID-3.5.167/BIOGRID-ORGANISM-3.5.167.tab2.zip"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
338
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
339 #download zip file
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
340 r = requests.get(tab2_link)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
341 with open("BioGRID.zip", "wb") as code:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
342 code.write(r.content)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
343
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
344 #unzip files
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
345 with zipfile.ZipFile("BioGRID.zip", 'r') as zip_ref:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
346 if not os.path.exists("tmp_BioGRID"): os.makedirs("tmp_BioGRID")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
347 zip_ref.extractall("tmp_BioGRID")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
348
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
349 #import file of interest and build dictionary
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
350 file_path="tmp_BioGRID/BIOGRID-ORGANISM-"+species_dict[species].replace(" ","_")+"-3.5.167.tab2.txt"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
351 with open(file_path,"r") as handle :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
352 tab_file = csv.reader(handle,delimiter="\t")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
353 dico_network = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
354 GeneID_index=1
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
355 network_cols=[1,2,7,8,11,12,14,18,20]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
356 for line in tab_file :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
357 if line[GeneID_index] not in dico_network:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
358 dico_network[line[GeneID_index]]=[[line[i] for i in network_cols]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
359 else:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
360 dico_network[line[GeneID_index]].append([line[i] for i in network_cols])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
361
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
362 #delete tmp_BioGRID directory
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
363 os.remove("BioGRID.zip")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
364 shutil.rmtree("tmp_BioGRID", ignore_errors=True)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
365
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
366 #download NCBI2Reactome.txt file and build dictionary
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
367 with requests.Session() as s:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
368 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
369 r.encoding ="utf-8"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
370 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
371
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
372 dico_nodes = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
373 geneid_index=0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
374 pathway_description_index=3
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
375 species_index=5
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
376 for line in tab_file :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
377 if line[species_index]==species_dict[species]:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
378 if line[geneid_index] in dico_nodes :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
379 dico_nodes[line[geneid_index]].append(line[pathway_description_index])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
380 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
381 dico_nodes[line[geneid_index]] = [line[pathway_description_index]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
382
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
383 dico={}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
384 dico['network']=dico_network
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
385 dico['nodes']=dico_nodes
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
386
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
387 ##Bioplex
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
388 elif interactome=="bioplex":
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
389
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
390 with requests.Session() as s:
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
391 r = s.get('http://bioplex.hms.harvard.edu/data/BioPlex_interactionList_v4a.tsv')
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
392 r = r.content.decode('utf-8')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
393 bioplex = csv.reader(r.splitlines(), delimiter='\t')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
394
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
395 dico_network = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
396 dico_network["GeneID"]={}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
397 network_geneid_cols=[0,1,4,5,8]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
398 dico_network["UniProt-AC"]={}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
399 network_uniprot_cols=[2,3,4,5,8]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
400 dico_GeneID_to_UniProt = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
401 for line in bioplex :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
402 if line[0] not in dico_network["GeneID"]:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
403 dico_network["GeneID"][line[0]]=[[line[i] for i in network_geneid_cols]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
404 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
405 dico_network["GeneID"][line[0]].append([line[i] for i in network_geneid_cols])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
406 if line[1] not in dico_network["UniProt-AC"]:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
407 dico_network["UniProt-AC"][line[2]]=[[line[i] for i in network_uniprot_cols]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
408 else:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
409 dico_network["UniProt-AC"][line[2]].append([line[i] for i in network_uniprot_cols])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
410 dico_GeneID_to_UniProt[line[0]]=line[2]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
411
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
412 with requests.Session() as s:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
413 r = s.get('https://reactome.org/download/current/UniProt2Reactome.txt')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
414 r.encoding ="utf-8"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
415 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
416
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
417 dico_nodes_uniprot = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
418 uniProt_index=0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
419 pathway_description_index=3
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
420 species_index=5
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
421 for line in tab_file :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
422 if line[species_index]==species_dict[species]:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
423 if line[uniProt_index] in dico_nodes_uniprot :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
424 dico_nodes_uniprot[line[uniProt_index]].append(line[pathway_description_index])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
425 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
426 dico_nodes_uniprot[line[uniProt_index]] = [line[pathway_description_index]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
427
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
428 with requests.Session() as s:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
429 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
430 r.encoding ="utf-8"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
431 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
432
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
433 dico_nodes_geneid = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
434 geneid_index=0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
435 pathway_description_index=3
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
436 species_index=5
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
437 for line in tab_file :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
438 if line[species_index]==species_dict[species]:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
439 if line[geneid_index] in dico_nodes_geneid :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
440 dico_nodes_geneid[line[geneid_index]].append(line[pathway_description_index])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
441 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
442 dico_nodes_geneid[line[geneid_index]] = [line[pathway_description_index]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
443
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
444 dico={}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
445 dico_nodes={}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
446 dico_nodes['GeneID']=dico_nodes_geneid
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
447 dico_nodes['UniProt-AC']=dico_nodes_uniprot
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
448 dico['network']=dico_network
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
449 dico['nodes']=dico_nodes
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
450 dico['convert']=dico_GeneID_to_UniProt
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
451
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
452 ##Humap
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
453 elif interactome=="humap":
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
454
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
455 with requests.Session() as s:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
456 r = s.get('http://proteincomplexes.org/static/downloads/nodeTable.txt')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
457 r = r.content.decode('utf-8')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
458 humap_nodes = csv.reader(r.splitlines(), delimiter=',')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
459
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
460 dico_geneid_to_gene_name={}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
461 dico_protein_name={}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
462 for line in humap_nodes :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
463 if check_entrez_geneid(line[4]):
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
464 if line[4] not in dico_geneid_to_gene_name:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
465 dico_geneid_to_gene_name[line[4]]=line[3]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
466 if line[4] not in dico_protein_name:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
467 dico_protein_name[line[4]]=line[5]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
468
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
469 with requests.Session() as s:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
470 r = s.get('http://proteincomplexes.org/static/downloads/pairsWprob.txt')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
471 r = r.content.decode('utf-8')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
472 humap = csv.reader(r.splitlines(), delimiter='\t')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
473
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
474 dico_network = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
475 for line in humap :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
476 if check_entrez_geneid(line[0]) and check_entrez_geneid(line[1]):
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
477
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
478 interactant_A, interactant_B = get_interactant_name(line,dico_geneid_to_gene_name)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
479
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
480 #first interactant (first column)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
481 if line[0] not in dico_network:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
482 dico_network[line[0]]=[line[:2]+[interactant_A,interactant_B,line[2]]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
483 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
484 dico_network[line[0]].append(line[:2]+[interactant_A,interactant_B,line[2]])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
485
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
486 #second interactant (second column)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
487 if line[1] not in dico_network:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
488 dico_network[line[1]]=[[line[1],line[0],interactant_B,interactant_A,line[2]]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
489 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
490 dico_network[line[1]].append([line[1],line[0],interactant_B,interactant_A,line[2]])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
491
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
492 with requests.Session() as s:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
493 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
494 r.encoding ="utf-8"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
495 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
496
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
497 dico_nodes = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
498 geneid_index=0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
499 pathway_description_index=3
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
500 species_index=5
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
501 for line in tab_file :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
502 if line[species_index]==species_dict[species]:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
503 #Fill dictionary with pathways
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
504 if line[geneid_index] in dico_nodes :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
505 dico_nodes[line[geneid_index]].append(line[pathway_description_index])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
506 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
507 dico_nodes[line[geneid_index]] = [line[pathway_description_index]]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
508
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
509 dico={}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
510 dico['network']=dico_network
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
511 dico['nodes']=dico_nodes
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
512 dico['gene_name']=dico_geneid_to_gene_name
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
513 dico['protein_name']=dico_protein_name
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
514
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
515 #writing output
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
516 output_file = species+'_'+interactome+'_'+ time.strftime("%Y-%m-%d") + ".json"
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
517 path = os.path.join(target_directory,output_file)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
518 name = species+" ("+species_dict[species]+") "+time.strftime("%d/%m/%Y")
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
519 release = species+"_"+interactome+"_"+ time.strftime("%Y-%m-%d")
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
520 id = str(10000000000 - int(time.strftime("%Y%m%d")))
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
521
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
522 with open(path, 'w') as handle:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
523 json.dump(dico, handle, sort_keys=True)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
524
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
525 data_table_entry = dict(id=id, release=release, name = name, species = species, value = path)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
526 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_"+interactome+"_dictionaries")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
527
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
528 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
529 # 5. nextprot (add protein features)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
530 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
531
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
532 def Build_nextprot_ref_file(data_manager_dict,target_directory):
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
533 nextprot_ids_file = "nextprot_ac_list_all.txt"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
534 ids = id_list_from_nextprot_ftp(nextprot_ids_file,target_directory)
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
535
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
536 output_file = 'nextprot_ref_'+ time.strftime("%d-%m-%Y") + ".tsv"
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
537 path = os.path.join(target_directory,output_file)
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
538 name = "neXtProt release "+time.strftime("%d-%m-%Y")
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
539 release_id = "nextprot_ref_"+time.strftime("%d-%m-%Y")
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
540
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
541 output = open(path, 'w')
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
542 writer = csv.writer(output,delimiter="\t")
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
543
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
544 nextprot_file=[["NextprotID","MW","SeqLength","IsoPoint","Chr","SubcellLocations","Diseases","TMDomains","ProteinExistence"]]
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
545 writer.writerows(nextprot_file)
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
546
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
547 for id in ids :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
548 #print (id)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
549 query="https://api.nextprot.org/entry/"+id+".json"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
550 resp = requests.get(url=query)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
551 data = resp.json()
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
552
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
553 #get info from json dictionary
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
554 mass_mol = data["entry"]["isoforms"][0]["massAsString"]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
555 seq_length = data['entry']["isoforms"][0]["sequenceLength"]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
556 iso_elec_point = data['entry']["isoforms"][0]["isoelectricPointAsString"]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
557 chr_loc = data['entry']["chromosomalLocations"][0]["chromosome"]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
558 protein_existence = "PE"+str(data['entry']["overview"]['proteinExistence']['level'])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
559
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
560 #put all subcell loc in a set
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
561 if "subcellular-location" in data['entry']["annotationsByCategory"].keys() :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
562 subcell_locs = data['entry']["annotationsByCategory"]["subcellular-location"]
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
563 all_subcell_locs = set()
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
564 for loc in subcell_locs :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
565 all_subcell_locs.add(loc['cvTermName'])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
566 all_subcell_locs.discard("")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
567 all_subcell_locs = ";".join(all_subcell_locs)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
568 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
569 all_subcell_locs = "NA"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
570
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
571 #put all subcell loc in a set
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
572 if ('disease') in data['entry']['annotationsByCategory'].keys() :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
573 diseases = data['entry']['annotationsByCategory']['disease']
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
574 all_diseases = set()
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
575 for disease in diseases :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
576 if (disease['cvTermName'] is not None and disease['cvTermName'] != ""):
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
577 all_diseases.add(disease['cvTermName'])
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
578 if len(all_diseases) > 0 : all_diseases = ";".join(all_diseases)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
579 else : all_diseases="NA"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
580 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
581 all_diseases="NA"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
582
1
f3507260b30f planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 0
diff changeset
583 #get all tm domain
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
584 nb_domains = 0
1
f3507260b30f planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 0
diff changeset
585 if "transmembrane-region" in data['entry']['annotationsByCategory'].keys():
f3507260b30f planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 0
diff changeset
586 tm_domains = data['entry']['annotationsByCategory']["transmembrane-region"]
f3507260b30f planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 0
diff changeset
587 all_tm_domains = set()
f3507260b30f planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 0
diff changeset
588 for tm in tm_domains :
f3507260b30f planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 0
diff changeset
589 all_tm_domains.add(tm['cvTermName'])
f3507260b30f planemo upload commit 39a9e2bf22b07beeca3fb77d86cda25820eb309c
proteore
parents: 0
diff changeset
590 nb_domains+=1
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
591 #print "nb domains ++"
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
592 #print (nb_domains)
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
593 nextprot_file[:] = []
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
594 nextprot_file.append([id,mass_mol,str(seq_length),iso_elec_point,chr_loc,all_subcell_locs,all_diseases,str(nb_domains),protein_existence])
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
595 writer.writerows(nextprot_file)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
596
2
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
597 id = str(10000000000 - int(time.strftime("%Y%m%d")))
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
598
9ec42cb35abd planemo upload commit 339ab77a83db03409c8001324f10b36ff5b13a39-dirty
proteore
parents: 1
diff changeset
599 data_table_entry = dict(id=id, release=release_id, name = name, value = path)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
600 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_nextprot_ref")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
601
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
602 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
603 # Main function
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
604 #######################################################################################################
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
605 def main():
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
606 parser = argparse.ArgumentParser()
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
607 parser.add_argument("--hpa", metavar = ("HPA_OPTION"))
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
608 parser.add_argument("--peptideatlas", metavar=("SAMPLE_CATEGORY_ID"))
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
609 parser.add_argument("--id_mapping", metavar = ("ID_MAPPING_SPECIES"))
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
610 parser.add_argument("--interactome", metavar = ("PPI"))
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
611 parser.add_argument("--species")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
612 parser.add_argument("--date")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
613 parser.add_argument("-o", "--output")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
614 parser.add_argument("--database")
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
615 parser.add_argument("--tool_data_path")
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
616 args = parser.parse_args()
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
617
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
618 data_manager_dict = {}
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
619 # Extract json file params
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
620 filename = args.output
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
621 params = from_json_string(open(filename).read())
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
622 target_directory = params[ 'output_data' ][0]['extra_files_path']
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
623 os.mkdir(target_directory)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
624
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
625 ## Download source files from HPA
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
626 try:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
627 hpa = args.hpa
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
628 except NameError:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
629 hpa = None
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
630 if hpa is not None:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
631 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
632 hpa = hpa.split(",")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
633 for hpa_tissue in hpa:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
634 HPA_sources(data_manager_dict, hpa_tissue, target_directory)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
635
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
636 ## Download source file from Peptide Atlas query
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
637 try:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
638 peptide_atlas = args.peptideatlas
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
639 date = args.date
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
640 except NameError:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
641 peptide_atlas = None
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
642 if peptide_atlas is not None:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
643 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
644 peptide_atlas = peptide_atlas.split(",")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
645 for pa_tissue in peptide_atlas:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
646 peptide_atlas_sources(data_manager_dict, pa_tissue, date, target_directory)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
647
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
648 ## Download ID_mapping source file from Uniprot
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
649 try:
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
650 id_mapping = args.id_mapping
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
651 except NameError:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
652 id_mapping = None
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
653 if id_mapping is not None:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
654 id_mapping = id_mapping .split(",")
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
655 for species in id_mapping :
4
e967a99d66b3 "planemo upload commit 540dd383c0617193db43bf11457011888751b022-dirty"
proteore
parents: 3
diff changeset
656 id_mapping_sources(data_manager_dict, species, target_directory, args.tool_data_path)
0
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
657
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
658 ## Download PPI ref files from biogrid/bioplex/humap
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
659 try:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
660 interactome=args.interactome
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
661 if interactome == "biogrid" :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
662 species=args.species
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
663 else :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
664 species="Human"
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
665 except NameError:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
666 interactome=None
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
667 species=None
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
668 if interactome is not None and species is not None:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
669 PPI_ref_files(data_manager_dict, species, interactome, target_directory)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
670
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
671 ## Build nextprot ref file for add protein features
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
672 try:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
673 database=args.database
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
674 except NameError:
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
675 database=None
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
676 if database is not None :
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
677 Build_nextprot_ref_file(data_manager_dict,target_directory)
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
678
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
679 #save info to json file
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
680 filename = args.output
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
681 open(filename, 'wb').write(to_json_string(data_manager_dict))
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
682
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
683 if __name__ == "__main__":
9e31ea9fc7ea planemo upload commit 567ba7934c0ca55529dfeb5e7ca0935ace260ad7-dirty
proteore
parents:
diff changeset
684 main()