data_manager_proteore: data_manager/resource

annotate data_manager/resource_building.py @ 46:80fc0b28e227 draft

planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty

author	dchristiany
date	Fri, 01 Feb 2019 10:21:58 -0500
parents	3febf3d1139a
children	7b486b0fba4e

rev	line source
19 85532a48e4e4 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 17 diff changeset	1 # -- coding: utf-8 --
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	2 """
10 2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 7 diff changeset	3 The purpose of this script is to create source files from different databases to be used in other proteore tools
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	4 """
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	5
21 0a79066992fc planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 20 diff changeset	6 import os, sys, argparse, requests, time, csv, re, json, shutil, zipfile
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	7 from io import BytesIO
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	8 from zipfile import ZipFile
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	9 from galaxy.util.json import from_json_string, to_json_string
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	10
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	11 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	12 # General functions
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	13 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	14 def unzip(url, output_file):
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	15 """
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	16 Get a zip file content from a link and unzip
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	17 """
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	18 content = requests.get(url)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	19 zipfile = ZipFile(BytesIO(content.content))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	20 output_content = ""
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	21 output_content += zipfile.open(zipfile.namelist()[0]).read()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	22 output = open(output_file, "w")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	23 output.write(output_content)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	24 output.close()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	25
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	26 def _add_data_table_entry(data_manager_dict, data_table_entry,data_table):
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	27 data_manager_dict['data_tables'] = data_manager_dict.get('data_tables', {})
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	28 data_manager_dict['data_tables'][data_table] = data_manager_dict['data_tables'].get(data_table, [])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	29 data_manager_dict['data_tables'][data_table].append(data_table_entry)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	30 return data_manager_dict
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	31
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	32 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	33 # 1. Human Protein Atlas
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	34 # - Normal tissue
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	35 # - Pathology
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	36 # - Full Atlas
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	37 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	38 def HPA_sources(data_manager_dict, tissue, target_directory):
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	39 if tissue == "HPA_normal_tissue":
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	40 tissue_name = "HPA normal tissue"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	41 url = "https://www.proteinatlas.org/download/normal_tissue.tsv.zip"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	42 elif tissue == "HPA_pathology":
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	43 tissue_name = "HPA pathology"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	44 url = "https://www.proteinatlas.org/download/pathology.tsv.zip"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	45 elif tissue == "HPA_full_atlas":
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	46 tissue_name = "HPA full atlas"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	47 url = "https://www.proteinatlas.org/download/proteinatlas.tsv.zip"
10 2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 7 diff changeset	48
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	49 output_file = tissue +"_"+ time.strftime("%d-%m-%Y") + ".tsv"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	50 path = os.path.join(target_directory, output_file)
10 2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 7 diff changeset	51 unzip(url, path) #download and save file
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	52 tissue_name = tissue_name + " " + time.strftime("%d/%m/%Y")
10 2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 7 diff changeset	53 tissue_id = tissue_name.replace(" ","_").replace("/","-")
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 7 diff changeset	54
2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 7 diff changeset	55 data_table_entry = dict(id=tissue_id, name = tissue_name, value = tissue, path = path)
12 60cb0a5ae661 planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 10 diff changeset	56 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_protein_atlas")
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	57
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	58
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	59 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	60 # 2. Peptide Atlas
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	61 #######################################################################################################
43 3febf3d1139a planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty dchristiany parents: 42 diff changeset	62 def peptide_atlas_sources(data_manager_dict, tissue, date, target_directory):
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	63 # Define organism_id (here Human) - to be upraded when other organism added to the project
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	64 organism_id = "2"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	65 # Extract sample_category_id and output filename
34 0c0586ac3e29 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 33 diff changeset	66 tissue=tissue.split(".")
29 871a7347ca24 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 28 diff changeset	67 sample_category_id = tissue[0]
34 0c0586ac3e29 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 33 diff changeset	68 tissue_name = tissue[1]
43 3febf3d1139a planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty dchristiany parents: 42 diff changeset	69 output_file = tissue_name+"_"+date + ".tsv"
30 b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	70
b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	71 query="https://db.systemsbiology.net/sbeams/cgi/PeptideAtlas/GetProteins?&atlas_build_id="+ \
b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	72 sample_category_id+"&display_options=ShowAbundances&organism_id="+organism_id+ \
b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	73 "&redundancy_constraint=4&presence_level_constraint=1%2C2&gene_annotation_level_constraint=leaf\
b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	74 &QUERY_NAME=AT_GetProteins&action=QUERY&output_mode=tsv&apply_action=QUERY"
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	75
30 b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	76 with requests.Session() as s:
b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	77 download = s.get(query)
b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	78 decoded_content = download.content.decode('utf-8')
b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	79 cr = csv.reader(decoded_content.splitlines(), delimiter='\t')
b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	80
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	81 uni_dict = build_dictionary(cr)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	82
10 2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 7 diff changeset	83 #columns of data table peptide_atlas
34 0c0586ac3e29 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 33 diff changeset	84 tissue_id = tissue_name+"_"+date
0c0586ac3e29 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 33 diff changeset	85 name = tissue_id.replace("-","/").replace("_"," ")
36 a4811c440b45 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 35 diff changeset	86 path = os.path.join(target_directory,output_file)
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	87
30 b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	88 with open(path,"w") as out :
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	89 w = csv.writer(out,delimiter='\t')
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	90 w.writerow(["Uniprot_AC","nb_obs"])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	91 w.writerows(uni_dict.items())
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	92
34 0c0586ac3e29 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 33 diff changeset	93 data_table_entry = dict(id=tissue_id, name=name, value = path, tissue = tissue_name)
12 60cb0a5ae661 planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 10 diff changeset	94 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_peptide_atlas")
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	95
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	96 #function to count the number of observations by uniprot id
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	97 def build_dictionary (csv) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	98 uni_dict = {}
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	99 for line in csv :
30 b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	100 if "-" not in line[0] and check_uniprot_access(line[0]) :
b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	101 if line[0] in uni_dict :
b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	102 uni_dict[line[0]] += int(line[5])
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	103 else :
30 b8271b9a1049 planemo upload commit c89c5deac442c0c2aa52b24f2c5af4b290773fc0-dirty dchristiany parents: 29 diff changeset	104 uni_dict[line[0]] = int(line[5])
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	105
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	106 return uni_dict
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	107
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	108 #function to check if an id is an uniprot accession number : return True or False-
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	109 def check_uniprot_access (id) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	110 uniprot_pattern = re.compile("[OPQ][0-9][A-Z0-9]{3}[0-9]\|[A-NR-Z][0-9]([A-Z][A-Z0-9]{2}[0-9]){1,2}")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	111 if uniprot_pattern.match(id) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	112 return True
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	113 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	114 return False
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	115
46 80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	116 def check_entrez_geneid (id) :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	117 entrez_pattern = re.complie("[0-9]+\|[A-Z]{1,2}_[0-9]+\|[A-Z]{1,2}_[A-Z]{1,4}[0-9]+")
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	118 if entrez_pattern.match(id) :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	119 return True
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	120 else :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	121 return False
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	122
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	123 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	124 # 3. ID mapping file
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	125 #######################################################################################################
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	126 import ftplib, gzip
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	127 csv.field_size_limit(sys.maxsize) # to handle big files
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	128
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	129 def id_mapping_sources (data_manager_dict, species, target_directory) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	130
28 d235909789ca planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty dchristiany parents: 27 diff changeset	131 human = species == "Human"
d235909789ca planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty dchristiany parents: 27 diff changeset	132 species_dict = { "Human" : "HUMAN_9606", "Mouse" : "MOUSE_10090", "Rat" : "RAT_10116" }
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	133 files=["idmapping_selected.tab.gz","idmapping.dat.gz"]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	134
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	135 #header
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	136 if human : tab = [["UniProt-AC","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","neXtProt","BioGrid","STRING","KEGG"]]
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	137 else : tab = [["UniProt-AC","UniProt-ID","GeneID","RefSeq","GI","PDB","GO","PIR","MIM","UniGene","Ensembl_Gene","Ensembl_Transcript","Ensembl_Protein","BioGrid","STRING","KEGG"]]
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	138
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	139 #print("header ok")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	140
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	141 #get selected.tab and keep only ids of interest
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	142 selected_tab_file=species_dict[species]+"_"+files[0]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	143 tab_path = download_from_uniprot_ftp(selected_tab_file,target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	144 with gzip.open(tab_path,"rt") as select :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	145 tab_reader = csv.reader(select,delimiter="\t")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	146 for line in tab_reader :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	147 tab.append([line[i] for i in [0,1,2,3,4,5,6,11,13,14,18,19,20]])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	148 os.remove(tab_path)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	149
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	150 #print("selected_tab ok")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	151
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	152 """
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	153 Supplementary ID to get from HUMAN_9606_idmapping.dat :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	154 -NextProt,BioGrid,STRING,KEGG
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	155 """
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	156
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	157 #there's more id type for human
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	158 if human : ids = ['neXtProt','BioGrid','STRING','KEGG' ] #ids to get from dat_file
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	159 else : ids = ['BioGrid','STRING','KEGG' ]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	160 unidict = {}
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	161
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	162 #keep only ids of interest in dictionaries
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	163 dat_file=species_dict[species]+"_"+files[1]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	164 dat_path = download_from_uniprot_ftp(dat_file,target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	165 with gzip.open(dat_path,"rt") as dat :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	166 dat_reader = csv.reader(dat,delimiter="\t")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	167 for line in dat_reader :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	168 uniprotID=line[0] #UniProtID as key
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	169 id_type=line[1] #ID type of corresponding id, key of sub-dictionnary
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	170 cor_id=line[2] #corresponding id
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	171 if "-" not in id_type : #we don't keep isoform
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	172 if id_type in ids and uniprotID in unidict :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	173 if id_type in unidict[uniprotID] :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	174 unidict[uniprotID][id_type]= ";".join([unidict[uniprotID][id_type],cor_id]) #if there is already a value in the dictionnary
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	175 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	176 unidict[uniprotID].update({ id_type : cor_id })
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	177 elif id_type in ids :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	178 unidict[uniprotID]={id_type : cor_id}
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	179 os.remove(dat_path)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	180
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	181 #print("dat_file ok")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	182
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	183 #add ids from idmapping.dat to the final tab
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	184 for line in tab[1:] :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	185 uniprotID=line[0]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	186 if human :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	187 if uniprotID in unidict :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	188 nextprot = access_dictionary(unidict,uniprotID,'neXtProt')
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	189 if nextprot != '' : nextprot = clean_nextprot_id(nextprot,line[0])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	190 line.extend([nextprot,access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'),
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	191 access_dictionary(unidict,uniprotID,'KEGG')])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	192 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	193 line.extend(["","","",""])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	194 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	195 if uniprotID in unidict :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	196 line.extend([access_dictionary(unidict,uniprotID,'BioGrid'),access_dictionary(unidict,uniprotID,'STRING'),
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	197 access_dictionary(unidict,uniprotID,'KEGG')])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	198 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	199 line.extend(["","",""])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	200
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	201 #print ("tab ok")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	202
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	203 #add missing nextprot ID for human
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	204 if human :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	205 #build next_dict
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	206 nextprot_ids = id_list_from_nextprot_ftp("nextprot_ac_list_all.txt",target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	207 next_dict = {}
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	208 for nextid in nextprot_ids :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	209 next_dict[nextid.replace("NX_","")] = nextid
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	210 os.remove(os.path.join(target_directory,"nextprot_ac_list_all.txt"))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	211
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	212 #add missing nextprot ID
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	213 for line in tab[1:] :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	214 uniprotID=line[0]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	215 nextprotID=line[13]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	216 if nextprotID == '' and uniprotID in next_dict :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	217 line[13]=next_dict[uniprotID]
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	218
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	219 output_file = species+"_id_mapping_"+ time.strftime("%d-%m-%Y") + ".tsv"
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	220 path = os.path.join(target_directory,output_file)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	221
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	222 with open(path,"w") as out :
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	223 w = csv.writer(out,delimiter='\t')
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	224 w.writerows(tab)
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	225
28 d235909789ca planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty dchristiany parents: 27 diff changeset	226 name_dict={"Human" : "Homo sapiens", "Mouse" : "Mus musculus", "Rat" : "Rattus norvegicus"}
d235909789ca planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty dchristiany parents: 27 diff changeset	227 name = species +" (" + name_dict[species]+" "+time.strftime("%d/%m/%Y")+")"
10 2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 7 diff changeset	228 id = species+"_id_mapping_"+ time.strftime("%d-%m-%Y")
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	229
10 2f153b41b6fe planemo upload commit e5e768b479ddc6b36270a1b5b0443a4c80d693bc-dirty dchristiany parents: 7 diff changeset	230 data_table_entry = dict(id=id, name = name, value = species, path = path)
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	231 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_id_mapping")
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	232
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	233 def download_from_uniprot_ftp(file,target_directory) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	234 ftp_dir = "pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	235 path = os.path.join(target_directory, file)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	236 ftp = ftplib.FTP("ftp.uniprot.org")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	237 ftp.login("anonymous", "anonymous")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	238 ftp.cwd(ftp_dir)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	239 ftp.retrbinary("RETR " + file, open(path, 'wb').write)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	240 ftp.quit()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	241 return (path)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	242
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	243 def id_list_from_nextprot_ftp(file,target_directory) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	244 ftp_dir = "pub/current_release/ac_lists/"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	245 path = os.path.join(target_directory, file)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	246 ftp = ftplib.FTP("ftp.nextprot.org")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	247 ftp.login("anonymous", "anonymous")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	248 ftp.cwd(ftp_dir)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	249 ftp.retrbinary("RETR " + file, open(path, 'wb').write)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	250 ftp.quit()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	251 with open(path,'r') as nextprot_ids :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	252 nextprot_ids = nextprot_ids.read().splitlines()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	253 return (nextprot_ids)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	254
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	255 #return '' if there's no value in a dictionary, avoid error
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	256 def access_dictionary (dico,key1,key2) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	257 if key1 in dico :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	258 if key2 in dico[key1] :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	259 return (dico[key1][key2])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	260 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	261 return ("")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	262 #print (key2,"not in ",dico,"[",key1,"]")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	263 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	264 return ('')
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	265
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	266 #if there are several nextprot ID for one uniprotID, return the uniprot like ID
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	267 def clean_nextprot_id (next_id,uniprotAc) :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	268 if len(next_id.split(";")) > 1 :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	269 tmp = next_id.split(";")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	270 if "NX_"+uniprotAc in tmp :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	271 return ("NX_"+uniprotAc)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	272 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	273 return (tmp[1])
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	274 else :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	275 return (next_id)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	276
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	277
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	278 #######################################################################################################
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	279 # 4. Build protein interaction maps files
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	280 #######################################################################################################
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	281
46 80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	282 def get_interactant_name(line):
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	283
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	284 if line[0] in dico_geneid_to_gene_name :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	285 print line[0]
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	286 interactant_A = dico_geneid_to_gene_name[line[0]]
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	287 else :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	288 interactant_A = "NA"
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	289
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	290 if line[1] in dico_geneid_to_gene_name :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	291 interactant_B = dico_geneid_to_gene_name[line[1]]
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	292 else :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	293 interactant_B = "NA"
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	294
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	295 return interactant_A, interactant_B
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	296
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	297 def PPI_ref_files(data_manager_dict, species, interactome, target_directory):
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	298
28 d235909789ca planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty dchristiany parents: 27 diff changeset	299 species_dict={'Human':'Homo sapiens',"Mouse":"Mus musculus","Rat":"Rattus norvegicus"}
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	300
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	301 ##BioGRID
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	302 if interactome=="biogrid":
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	303
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	304 tab2_link="https://downloads.thebiogrid.org/Download/BioGRID/Release-Archive/BIOGRID-3.5.167/BIOGRID-ORGANISM-3.5.167.tab2.zip"
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	305
39 ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	306 #download zip file
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	307 r = requests.get(tab2_link)
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	308 with open("BioGRID.zip", "wb") as code:
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	309 code.write(r.content)
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	310
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	311 #unzip files
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	312 with zipfile.ZipFile("BioGRID.zip", 'r') as zip_ref:
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	313 if not os.path.exists("tmp_BioGRID"): os.makedirs("tmp_BioGRID")
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	314 zip_ref.extractall("tmp_BioGRID")
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	315
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	316 #import file of interest and build dictionary
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	317 file_path="tmp_BioGRID/BIOGRID-ORGANISM-"+species_dict[species].replace(" ","_")+"-3.5.167.tab2.txt"
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	318 with open(file_path,"r") as handle :
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	319 tab_file = csv.reader(handle,delimiter="\t")
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	320 dico_network = {}
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	321 GeneID_index=1
39 ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	322 network_cols=[1,2,7,8,11,12,14,18,20]
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	323 for line in tab_file :
22 778cc97cb115 planemo upload commit e2e4ac8c2c4dc5675d4c415ff192a925ca5e6b98-dirty dchristiany parents: 21 diff changeset	324 if line[GeneID_index] not in dico_network:
778cc97cb115 planemo upload commit e2e4ac8c2c4dc5675d4c415ff192a925ca5e6b98-dirty dchristiany parents: 21 diff changeset	325 dico_network[line[GeneID_index]]=[[line[i] for i in network_cols]]
778cc97cb115 planemo upload commit e2e4ac8c2c4dc5675d4c415ff192a925ca5e6b98-dirty dchristiany parents: 21 diff changeset	326 else:
778cc97cb115 planemo upload commit e2e4ac8c2c4dc5675d4c415ff192a925ca5e6b98-dirty dchristiany parents: 21 diff changeset	327 dico_network[line[GeneID_index]].append([line[i] for i in network_cols])
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	328
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	329 #delete tmp_BioGRID directory
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	330 os.remove("BioGRID.zip")
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	331 shutil.rmtree("tmp_BioGRID", ignore_errors=True)
39 ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	332
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	333 #download NCBI2Reactome.txt file and build dictionary
39 ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	334 with requests.Session() as s:
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	335 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt')
40 fddf4a3847f4 planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 39 diff changeset	336 r.encoding ="utf-8"
fddf4a3847f4 planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 39 diff changeset	337 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	338
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	339 dico_nodes = {}
46 80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	340 geneid_index=0
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	341 pathway_description_index=3
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	342 species_index=5
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	343 for line in tab_file :
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	344 if line[species_index]==species_dict[species]:
46 80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	345 if line[geneid_index] in dico_nodes :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	346 dico_nodes[line[geneid_index]].append(line[pathway_description_index])
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	347 else :
46 80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	348 dico_nodes[line[geneid_index]] = [line[pathway_description_index]]
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	349
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	350 dico={}
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	351 dico['network']=dico_network
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	352 dico['nodes']=dico_nodes
39 ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	353
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	354 ##Bioplex
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	355 elif interactome=="bioplex":
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	356
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	357 with requests.Session() as s:
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	358 r = s.get('http://bioplex.hms.harvard.edu/data/BioPlex_interactionList_v4a.tsv')
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	359 r = r.content.decode('utf-8')
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	360 bioplex = csv.reader(r.splitlines(), delimiter='\t')
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	361
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	362 dico_network = {}
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	363 dico_network["GeneID"]={}
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	364 network_geneid_cols=[0,1,4,5,8]
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	365 dico_network["UniProt-AC"]={}
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	366 network_uniprot_cols=[2,3,4,5,8]
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	367 dico_GeneID_to_UniProt = {}
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	368 for line in bioplex :
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	369 if line[0] not in dico_network["GeneID"]:
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	370 dico_network["GeneID"][line[0]]=[[line[i] for i in network_geneid_cols]]
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	371 else :
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	372 dico_network["GeneID"][line[0]].append([line[i] for i in network_geneid_cols])
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	373 if line[1] not in dico_network["UniProt-AC"]:
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	374 dico_network["UniProt-AC"][line[2]]=[[line[i] for i in network_uniprot_cols]]
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	375 else:
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	376 dico_network["UniProt-AC"][line[2]].append([line[i] for i in network_uniprot_cols])
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	377 dico_GeneID_to_UniProt[line[0]]=line[2]
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	378
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	379 with requests.Session() as s:
40 fddf4a3847f4 planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 39 diff changeset	380 r = s.get('https://reactome.org/download/current/UniProt2Reactome.txt')
fddf4a3847f4 planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 39 diff changeset	381 r.encoding ="utf-8"
fddf4a3847f4 planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 39 diff changeset	382 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
39 ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	383
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	384 dico_nodes_uniprot = {}
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	385 uniProt_index=0
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	386 pathway_description_index=3
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	387 species_index=5
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	388 for line in tab_file :
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	389 if line[species_index]==species_dict[species]:
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	390 if line[uniProt_index] in dico_nodes_uniprot :
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	391 dico_nodes_uniprot[line[uniProt_index]].append(line[pathway_description_index])
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	392 else :
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	393 dico_nodes_uniprot[line[uniProt_index]] = [line[pathway_description_index]]
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	394
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	395 with requests.Session() as s:
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	396 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt')
40 fddf4a3847f4 planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 39 diff changeset	397 r.encoding ="utf-8"
fddf4a3847f4 planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 39 diff changeset	398 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
39 ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	399
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	400 dico_nodes_geneid = {}
46 80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	401 geneid_index=0
39 ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	402 pathway_description_index=3
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	403 species_index=5
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	404 for line in tab_file :
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	405 if line[species_index]==species_dict[species]:
46 80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	406 if line[geneid_index] in dico_nodes_geneid :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	407 dico_nodes_geneid[line[geneid_index]].append(line[pathway_description_index])
39 ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	408 else :
46 80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	409 dico_nodes_geneid[line[geneid_index]] = [line[pathway_description_index]]
39 ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	410
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	411 dico={}
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	412 dico_nodes={}
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	413 dico_nodes['GeneID']=dico_nodes_geneid
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	414 dico_nodes['UniProt-AC']=dico_nodes_uniprot
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	415 dico['network']=dico_network
ec6252ad1a8e planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 36 diff changeset	416 dico['nodes']=dico_nodes
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	417 dico['convert']=dico_GeneID_to_UniProt
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	418
46 80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	419 ##Humap
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	420 elif interactome=="humap":
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	421
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	422 with requests.Session() as s:
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	423 r = s.get('http://proteincomplexes.org/static/downloads/nodeTable.txt')
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	424 r = r.content.decode('utf-8')
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	425 humap_nodes = csv.reader(r.splitlines(), delimiter=',')
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	426
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	427 dico_geneid_to_gene_name={}
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	428 for line in humap_nodes :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	429 if check_entrez_geneid(line[5]):
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	430 if line[5] not in dico_geneid_to_gene_name:
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	431 dico_geneid_to_gene_name[line[5]]=[line[4]]
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	432 else :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	433 if line[4] not in dico_geneid_to_gene_name[line[5]] :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	434 dico_geneid_to_gene_name[line[5]].append(line[4])
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	435
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	436 with requests.Session() as s:
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	437 r = s.get('http://proteincomplexes.org/static/downloads/pairsWprob.txt')
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	438 r = r.content.decode('utf-8')
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	439 humap = csv.reader(r.splitlines(), delimiter='\t')
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	440
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	441 dico_network = {}
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	442 for line in humap :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	443 if check_entrez_geneid(line[0]) and check_entrez_geneid(line[1]):
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	444
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	445 interactant_A, interactant_B = get_interactant_name(line,dico_geneid_to_gene_name)
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	446
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	447 if line[0] not in dico_network:
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	448 dico_network[line[0]]=[line[:2]+[interactant_A,interactant_B,line[2]]]
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	449 else :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	450 dico_network[line[0]].append(line[:2]+[interactant_A,interactant_B,line[2]])
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	451
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	452 with requests.Session() as s:
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	453 r = s.get('https://www.reactome.org/download/current/NCBI2Reactome.txt')
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	454 r.encoding ="utf-8"
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	455 tab_file = csv.reader(r.content.splitlines(), delimiter='\t')
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	456
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	457 dico_nodes = {}
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	458 geneid_index=0
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	459 pathway_description_index=3
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	460 species_index=5
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	461 for line in tab_file :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	462 if line[species_index]==species_dict[species]:
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	463 #Fill dictionary with pathways
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	464 if line[geneid_index] in dico_nodes :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	465 dico_nodes[line[geneid_index]].append(line[pathway_description_index])
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	466 else :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	467 dico_nodes[line[geneid_index]] = [line[pathway_description_index]]
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	468
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	469 dico={}
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	470 dico['network']=dico_network
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	471 dico['nodes']=dico_nodes
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	472 dico['gene_name']=dico_geneid_to_gene_name
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	473
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	474 #writing output
42 5a37a086c9a8 planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 41 diff changeset	475 output_file = species+'_'+interactome+'_'+ time.strftime("%d-%m-%Y") + ".json"
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	476 path = os.path.join(target_directory,output_file)
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	477 name = species+" ("+species_dict[species]+") "+time.strftime("%d/%m/%Y")
27 592c59530c32 planemo upload commit e694c4b0df30a4286ba09721696e8ec3af25fd97-dirty dchristiany parents: 26 diff changeset	478 id = species+"_"+interactome+"_"+ time.strftime("%d-%m-%Y")
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	479
41 4062484f2cee planemo upload commit 43e2a01d7519104c2c16510e4dbdc023e65c49c7-dirty dchristiany parents: 40 diff changeset	480 with open(path, 'w') as handle:
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	481 json.dump(dico, handle, sort_keys=True)
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	482
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	483 data_table_entry = dict(id=id, name = name, value = species, path = path)
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	484 _add_data_table_entry(data_manager_dict, data_table_entry, "proteore_"+interactome+"_dictionaries")
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	485
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	486
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	487 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	488 # Main function
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	489 #######################################################################################################
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	490 def main():
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	491 parser = argparse.ArgumentParser()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	492 parser.add_argument("--hpa", metavar = ("HPA_OPTION"))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	493 parser.add_argument("--peptideatlas", metavar=("SAMPLE_CATEGORY_ID"))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	494 parser.add_argument("--id_mapping", metavar = ("ID_MAPPING_SPECIES"))
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	495 parser.add_argument("--interactome", metavar = ("PPI"))
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	496 parser.add_argument("--species")
43 3febf3d1139a planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty dchristiany parents: 42 diff changeset	497 parser.add_argument("--date")
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	498 parser.add_argument("-o", "--output")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	499 args = parser.parse_args()
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	500
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	501 data_manager_dict = {}
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	502 # Extract json file params
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	503 filename = args.output
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	504 params = from_json_string(open(filename).read())
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	505 target_directory = params[ 'output_data' ][0]['extra_files_path']
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	506 os.mkdir(target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	507
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	508 ## Download source files from HPA
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	509 try:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	510 hpa = args.hpa
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	511 except NameError:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	512 hpa = None
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	513 if hpa is not None:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	514 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	515 hpa = hpa.split(",")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	516 for hpa_tissue in hpa:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	517 HPA_sources(data_manager_dict, hpa_tissue, target_directory)
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	518
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	519 ## Download source file from Peptide Atlas query
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	520 try:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	521 peptide_atlas = args.peptideatlas
43 3febf3d1139a planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty dchristiany parents: 42 diff changeset	522 date = args.date
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	523 except NameError:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	524 peptide_atlas = None
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	525 if peptide_atlas is not None:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	526 #target_directory = "/projet/galaxydev/galaxy/tools/proteore/ProteoRE/tools/resources_building/test-data/"
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	527 peptide_atlas = peptide_atlas.split(",")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	528 for pa_tissue in peptide_atlas:
43 3febf3d1139a planemo upload commit 9b701c1faf8be4835b4e7236780ee9ee26f9a373-dirty dchristiany parents: 42 diff changeset	529 peptide_atlas_sources(data_manager_dict, pa_tissue, date, target_directory)
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	530
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	531 ## Download ID_mapping source file from Uniprot
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	532 try:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	533 id_mapping=args.id_mapping
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	534 except NameError:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	535 id_mapping = None
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	536 if id_mapping is not None:
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	537 id_mapping = id_mapping .split(",")
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	538 for species in id_mapping :
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	539 id_mapping_sources(data_manager_dict, species, target_directory)
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	540
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	541 ## Download PPI ref files from biogrid/bioplex/humap
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	542 try:
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	543 interactome=args.interactome
46 80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	544 if interactome == "biogrid" :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	545 species=args.species
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	546 else :
80fc0b28e227 planemo upload commit cb633de1f04ef7b7133728909716b6c6594533d1-dirty dchristiany parents: 43 diff changeset	547 species="Human"
15 83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	548 except NameError:
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	549 interactome=None
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	550 species=None
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	551 if interactome is not None and species is not None:
83f57ba70416 planemo upload commit 8040003119a3d54866ec6ee9b9f659f2af554817-dirty dchristiany parents: 13 diff changeset	552 PPI_ref_files(data_manager_dict, species, interactome, target_directory)
0 2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	553
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	554 #save info to json file
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	555 filename = args.output
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	556 open(filename, 'wb').write(to_json_string(data_manager_dict))
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	557
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	558 if __name__ == "__main__":
2de84fea8367 planemo upload commit d703392579d96e480c6461ce679516b12cefb3de-dirty dchristiany parents: diff changeset	559 main()

Mercurial > repos > dchristiany > data_manager_proteore

annotate data_manager/resource_building.py @ 46:80fc0b28e227 draft