Mercurial > repos > infr > impc_tools

annotate impc_tool.py @ 0:0a9cf7f52b9c draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
author infr
date Wed, 22 Jun 2022 13:36:44 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
1 import sys
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
2 import requests
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
3 import pandas as pd
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
4 import urllib.request as url
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
5
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
6 impc_api_url = "https://www.gentar.org/impc-dev-api/"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
7 impc_api_search_url = f"{impc_api_url}/genes"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
8 impc_api_gene_bundle_url = f"{impc_api_url}/geneBundles"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
9
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
10
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
11 def stop_err(msg):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
12 sys.exit(msg)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
13
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
14
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
15 def main():
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
16 inp = str(sys.argv[1])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
17 query = str(sys.argv[3])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
18
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
19 try:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
20 if query == '7':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
21 full_gene_table()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
22 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
23
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
24 if str(sys.argv[5])=="txt":
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
25 s = str(sys.argv[6])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
26 if s == "t":
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
27 sep = "\t"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
28 elif s == "s":
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
29 sep = " "
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
30 elif s in ",;.":
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
31 sep = s
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
32 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
33 sys.exit("Separator not valid, please change it.")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
34 inp = pd.read_csv(inp, header=None, delimiter=sep)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
35 if len(inp.columns)==1:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
36 inp = str(inp[0].values[0]).replace("'","")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
37 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
38 inp = inp.to_string(header=False, index=False).replace(" ",",")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
39
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
40 if query == '8':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
41 genes_in_pipeline(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
42 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
43 elif query == '10': # it's here but not totally implemented
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
44 par_pip_ma(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
45 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
46 elif query == '11': # it's here but not totally implemented
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
47 par_gen(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
48 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
49 elif query == '2' or query == "4":
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
50 final_list=pheno_mapping(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
51 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
52 final_list=gene_mapping(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
53 inp= ",".join(final_list)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
54
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
55
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
56 if query == '1':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
57 get_pheno(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
58 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
59 elif query == '2':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
60 get_genes(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
61 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
62 elif query == '3':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
63 gene_set(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
64 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
65 elif query == '4':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
66 extr_img(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
67 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
68 elif query == '5':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
69 parameters(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
70 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
71 elif query == '6':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
72 sign_par(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
73 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
74 elif query == '9':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
75 sign_mp(inp)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
76 sys.exit(0)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
77 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
78 stop_err("Error, non-implemented query selected: " + query)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
79 except Exception as ex:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
80 stop_err('Error running impc_tool.py:\n' + str(ex))
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
81
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
82
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
83 # 1-Given a gene id, retrieve all the phenotypes related to it (id and name)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
84 def get_pheno(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
85 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
86 mgi_accession_id = inp
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
87
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
88 gene_url = f"{impc_api_search_url}/{mgi_accession_id}"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
89 gene_data = requests.get(gene_url).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
90
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
91 p_list = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
92 id_list = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
93
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
94 if gene_data['significantMpTerms'] == None:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
95 stop_err("No significant MP terms found for this gene")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
96 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
97 for x in gene_data['significantMpTerms']:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
98 p_list.append(x['mpTermId'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
99 id_list.append(x['mpTermName'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
100
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
101 df = pd.DataFrame()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
102 df['MP term name'] = p_list
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
103 df['MP term id'] = id_list
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
104
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
105 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
106 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
107 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
108 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
109
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
110
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
111 # 3-Extract all genes having a particular phenotype or a set of phenotypes (e.g. relevant to a disease)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
112 def get_genes(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
113 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
114 target_mp_terms = inp
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
115
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
116 ## All the data is paginated using the page and size parameters, by default the endpoint returns the first 20 hits
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
117 gene_by_phenotypes_query = f"{impc_api_search_url}/search/findAllBySignificantMpTermIdsContains?mpTermIds={target_mp_terms}&page=0&size=20"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
118 genes_with_clinical_chemistry_phenotypes = requests.get(gene_by_phenotypes_query).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
119 print(f"Genes with {target_mp_terms}: {genes_with_clinical_chemistry_phenotypes['page']['totalElements']}")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
120 list_of_genes = pd.DataFrame(columns=['Gene accession id', 'Gene name', 'Gene bundle url'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
121 acc = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
122 name = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
123 url = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
124
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
125 for gene in genes_with_clinical_chemistry_phenotypes['_embedded']['genes']:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
126 acc.append(gene['mgiAccessionId'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
127 name.append(gene['markerName'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
128 url.append(gene['_links']['geneBundle']['href'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
129
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
130 list_of_genes['Gene accession id'] = acc
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
131 list_of_genes['Gene name'] = name
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
132 list_of_genes['Gene bundle url'] = url
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
133
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
134 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
135 list_of_genes.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
136 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
137 list_of_genes.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
138
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
139 # 4. Extract all phenotypes which are present in a particular gene set (e.g. genes together in a pathway)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
140
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
141 def gene_set(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
142 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
143 target_genes = inp
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
144
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
145 genes_in_gene_list_query = f"{impc_api_search_url}/search/findAllByMgiAccessionIdIn?mgiAccessionIds={target_genes}"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
146
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
147 genes_in_gene_list = requests.get(genes_in_gene_list_query).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
148 list_of_mp_terms_vs_gene_index = {}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
149
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
150 for gene in genes_in_gene_list['_embedded']['genes']:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
151 mp_terms = gene['significantMpTerms']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
152 gene_acc_id = gene["mgiAccessionId"]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
153 if mp_terms is None:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
154 continue
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
155 for mp_term_name in mp_terms:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
156 if mp_term_name['mpTermId'] not in list_of_mp_terms_vs_gene_index:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
157 list_of_mp_terms_vs_gene_index[mp_term_name['mpTermId']] = {"mp_term": mp_term_name['mpTermId'], "mp_name": mp_term_name['mpTermName'], "genes": []}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
158 list_of_mp_terms_vs_gene_index[mp_term_name['mpTermId']]["genes"].append(gene_acc_id)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
159 genes_by_mp_term = list(list_of_mp_terms_vs_gene_index.values())
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
160
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
161 df = pd.DataFrame()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
162 terms = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
163 names = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
164 genes = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
165 for i in genes_by_mp_term:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
166 terms.append(i['mp_term'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
167 names.append(i['mp_name'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
168 genes.append(",".join(i['genes']))
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
169
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
170 df['mp_term'] = terms
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
171 df['mp_name'] = names
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
172 df['genes'] = genes
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
173
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
174 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
175 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
176 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
177 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
178
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
179 # 7. Extract images with a particular phenotype or a set of phenotypes
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
180
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
181
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
182 def extr_img(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
183 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
184 target_mp_terms = inp # ['MP:0002110', 'MP:0000559']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
185
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
186 ## All the data is paginated using the page and size parameters, by default the endpoint returns the first 20 hits
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
187 gene_by_phenotypes_query = f"{impc_api_search_url}/search/findAllBySignificantMpTermIdsContains?mpTermIds={target_mp_terms}&page=0&size=20"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
188 genes_with_morphology_mps = requests.get(gene_by_phenotypes_query).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
189 list_of_gene_bundle_urls = [gene["_links"]["geneBundle"]['href'] for gene in
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
190 genes_with_morphology_mps['_embedded']['genes']]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
191
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
192 gene_bundles = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
193 for gene_bundle_url in list_of_gene_bundle_urls:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
194 gene_bundle = requests.get(gene_bundle_url).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
195 gene_bundles.append(gene_bundle)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
196
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
197 images_with_morphology_mps = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
198
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
199 ## Doing just the first 20 and filtering out fields on the images
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
200 display_fields = ['geneSymbol', 'parameterName', 'biologicalSampleGroup', 'colonyId', 'zygosity', 'sex',
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
201 'downloadUrl', 'externalSampleId', 'thumbnailUrl']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
202
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
203
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
204 for gene_bundle in gene_bundles[:20]:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
205 if len(gene_bundle) == 4:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
206 continue
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
207 if gene_bundle["geneImages"] is not None:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
208 images = gene_bundle["geneImages"]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
209 for image in images:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
210 display_image = {k: v for k, v in image.items() if k in display_fields}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
211 images_with_morphology_mps.append(display_image)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
212
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
213 images_table = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
214 print(f"Images related to phenotype {target_mp_terms}: {len(images_with_morphology_mps)}")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
215 ## Displaying just the first 20 images
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
216 for i in images_with_morphology_mps[:20]:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
217 row = [f"<img src='{i['thumbnailUrl']}' />"] + list(i.values())
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
218 images_table.append(row)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
219
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
220 df = pd.DataFrame()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
221 externalSampleId = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
222 geneSymbol = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
223 biologicalSampleGroup = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
224 sex = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
225 colonyId = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
226 zygosity = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
227 parameterName = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
228 downloadUrl = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
229 thumbnailUrl = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
230
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
231 for i in images_table:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
232 externalSampleId.append(i[1])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
233 geneSymbol.append(i[2])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
234 biologicalSampleGroup.append(i[3])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
235 sex.append(i[4])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
236 colonyId.append(i[5])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
237 zygosity.append(i[6])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
238 parameterName.append(i[7])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
239 downloadUrl.append(i[8])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
240 thumbnailUrl.append(i[9])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
241
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
242 df['externalSampleId'] = externalSampleId
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
243 df['geneSymbol'] = geneSymbol
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
244 df['biologicalSampleGroup'] = biologicalSampleGroup
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
245 df['sex'] = sex
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
246 df['colonyId'] = colonyId
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
247 df['zygosity'] = zygosity
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
248 df['parameterName'] = parameterName
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
249 df['downloadUrl'] = downloadUrl
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
250 df['thumbnailUrl'] = thumbnailUrl
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
251
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
252 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
253 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
254 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
255 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
256
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
257 # 11- Which parameters have been measured for a particular knockout EASY
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
258
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
259
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
260 def parameters(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
261 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
262 knockout = inp # "MGI:104636"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
263 gene_info = requests.get(impc_api_search_url + "/" + knockout).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
264
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
265 if gene_info['phenotypingDataAvailable']:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
266 geneBundle = requests.get(gene_info['_links']['geneBundle']['href']).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
267 gen_imgs = geneBundle['geneImages']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
268 par_list = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
269 l = {}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
270 for i in gen_imgs:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
271 l = {"Parameter Name": i['parameterName']}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
272 if l not in par_list:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
273 par_list.append(l)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
274 df = pd.DataFrame()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
275 l = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
276
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
277 for i in par_list:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
278 l.append(i['Parameter Name'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
279
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
280 df['Parameter'] = l
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
281 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
282 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
283 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
284 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
285
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
286 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
287 stop_err("No parameters available for this knockout gene")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
288
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
289
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
290 # 12- Which parameters identified a significant finding for a particular knockout line (colony) EASY
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
291 def sign_par(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
292 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
293 knockout = inp # "MGI:104636"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
294
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
295 gene_info = requests.get(f"{impc_api_url}statisticalResults/search/findAllByMarkerAccessionIdIsAndSignificantTrue?mgiAccessionId=" + knockout).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
296 gene_stats = gene_info['_embedded']['statisticalResults']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
297
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
298 if len(gene_stats) == 0:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
299 stop_err("No statistically relevant parameters found for this knockout gene")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
300 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
301 df = pd.DataFrame()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
302 n = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
303 p = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
304 for g in gene_stats:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
305 n.append(g['parameterName'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
306 p.append(g['pvalue'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
307
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
308 df['Parameter name'] = n
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
309 df['p-value'] = p
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
310 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
311 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
312 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
313 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
314
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
315
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
316 # 13- List of genes names and ID measured in a pipeline
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
317 def genes_in_pipeline(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
318 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
319 pip = inp
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
320
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
321 g_in_p_query = f"{impc_api_search_url}/search/findAllByTestedPipelineId?pipelineId={pip}&page=0&size=1000"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
322 genes_in_pip = requests.get(g_in_p_query).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
323 pages = genes_in_pip['page']['totalPages']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
324 max_elem = genes_in_pip['page']['totalElements']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
325
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
326 print(f"Genes with {pip}: {genes_in_pip['page']['totalElements']}")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
327 d ={ }
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
328 list_d = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
329 list_of_genes = pd.DataFrame(columns=['Gene accession id', 'Gene name'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
330 acc = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
331 name = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
332
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
333 if max_elem > 1000:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
334 g_in_p_query = genes_in_pip['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
335 for i in range(1,pages):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
336 gl = requests.get(f'{impc_api_search_url}/search/findAllByTestedPipelineId?pipelineId={pip}&page={i}&size=1000').json()['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
337 g_in_p_query += gl
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
338 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
339 g_in_p_query = genes_in_pip['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
340
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
341 for g in g_in_p_query:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
342 d = {"Gene Accession ID": g['mgiAccessionId'], "Gene Name": g['markerName']}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
343 list_d.append(d)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
344
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
345 for i in list_d:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
346 acc.append(i['Gene Accession ID'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
347 name.append(i['Gene Name'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
348
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
349 list_of_genes['Gene accession id'] = acc
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
350 list_of_genes['Gene name'] = name
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
351
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
352 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
353 list_of_genes.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
354 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
355 list_of_genes.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
356
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
357
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
358 # 14- Extract all genes and corresponding phenotypes related to a particular organ system(eg: significatMPTerm)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
359 def sign_mp(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
360 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
361 mp_term = inp # ['MP:0005391']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
362
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
363 gene_by_mpterm_query = f"{impc_api_search_url}/search/findAllBySignificantMpTermIdsContains?mpTermIds={mp_term}&size=1000"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
364 genes_with_mpterm = requests.get(gene_by_mpterm_query).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
365
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
366 pages = genes_with_mpterm['page']['totalPages']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
367 genes_info = genes_with_mpterm['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
368
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
369 for pn in range(1,pages):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
370 pq = f"{impc_api_search_url}/search/findAllBySignificantMpTermIdsContains?mpTermIds={mp_term}&page={pn}&size=1000"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
371 g = requests.get(pq).json()['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
372 genes_info += g
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
373
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
374 list_d=[]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
375 d={}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
376 for g in genes_info:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
377 names=[]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
378 ids=[]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
379 for s in g['significantMpTerms']:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
380 names.append(s['mpTermName'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
381 ids.append(s['mpTermId'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
382 d={'Gene':g['mgiAccessionId'], 'mpTermId': ids, 'mpTermName':names}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
383 list_d.append(d)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
384
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
385
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
386 g = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
387 ids = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
388 names = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
389 for i in list_d:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
390 g.append(i['Gene'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
391 ids.append(i['mpTermId'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
392 names.append(i['mpTermName'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
393
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
394 df = pd.DataFrame()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
395 df['Gene Id']=g
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
396 df['Significant MP terms Ids']=ids
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
397 df['Significant MP terms Names']=names
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
398
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
399 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
400 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
401 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
402 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
403
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
404
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
405 # 16- Full table of genes and all identified phenotypes
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
406
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
407 def full_gene_table():
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
408 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
409 gene_list = requests.get(impc_api_search_url + '?page=0&size=1000').json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
410 pages = gene_list['page']['totalPages']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
411 genes_info = gene_list['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
412
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
413 for pn in range(1,pages):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
414 gp = requests.get(impc_api_search_url + f'?page={pn}&size=1000').json()['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
415 genes_info += gp
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
416
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
417 d = {}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
418 list_d=[]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
419
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
420 for i in genes_info:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
421 l = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
422 if i['significantMpTerms'] is None:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
423 d={"Gene": i['mgiAccessionId'], "Identified phenotypes": "None"}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
424 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
425 d = {"Gene": i['mgiAccessionId'], "Identified phenotypes": [sub['mpTermId'] for sub in i['significantMpTerms']]}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
426 list_d.append(d)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
427
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
428 df = pd.DataFrame()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
429 g = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
430 p = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
431 for i in list_d:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
432 g.append(i['Gene'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
433 p.append(i['Identified phenotypes'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
434
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
435 df['MGI id'] = g
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
436 df['MP term list'] = p
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
437
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
438 for i in range(0, len(df)):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
439 if df['MP term list'][i] != "None":
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
440 df['MP term list'][i] = str(df['MP term list'][i])[1:-1].replace("'", "")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
441
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
442 if str(sys.argv[1]) == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
443 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
444 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
445 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
446 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
447 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
448 df = df[df['MP term list'] != "None"]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
449 df.reset_index(drop=True, inplace=True)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
450 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
451 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
452 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
453 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
454
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
455 # Old method, chech which is faster
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
456 # max_elem = gene_list['page']['totalElements']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
457 # d = {}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
458 # list_d = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
459 # for i in range(0, pages):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
460 # gl = requests.get(impc_api_search_url + '?page=' + str(i) + '&size=' + str(max_elem)).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
461 # for g in gl['_embedded']['genes']:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
462 # if g['significantMpTerms'] is None:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
463 # d = {"Gene": g['mgiAccessionId'], "Identified phenotypes": "None"}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
464 # else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
465 # d = {"Gene": g['mgiAccessionId'], "Identified phenotypes": [ sub['mpTermId'] for sub in g['significantMpTerms'] ]}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
466 # list_d.append(d)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
467
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
468
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
469
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
470
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
471 # 18- Extract measurements and analysis for a parameter or pipeline
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
472
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
473 def par_pip_ma(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
474 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
475 id = inp
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
476
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
477 if id[0:4] == "IMPC":
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
478 par = True
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
479 ma_query = f"{impc_api_search_url}/search/findAllByTestedParameterId?parameterId={id}&page=0&size=1000"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
480 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
481 ma_query = f"{impc_api_search_url}/search/findAllByTestedPipelineId?pipelineId={id}&page=0&size=1000"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
482 par = False
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
483
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
484 ma_in_pip = requests.get(ma_query).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
485 pages = ma_in_pip['page']['totalPages']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
486 max_elem = ma_in_pip['page']['totalElements']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
487
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
488 print(f"Genes with {id}: {ma_in_pip['page']['totalElements']}")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
489 d = {}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
490 list_d = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
491 list_of_genes = pd.DataFrame(columns=['Measurements', 'Analysis'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
492 mes = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
493 an = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
494
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
495 if max_elem > 1000:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
496
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
497 ma_in_pip = ma_in_pip['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
498 for pn in range(1, pages):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
499 if par:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
500 pip = requests.get(f"{impc_api_search_url}/search/findAllByTestedParameterId?parameterId={id}&page={pn}&size=1000").json()['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
501 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
502 pip = requests.get(f"{impc_api_search_url}/search/findAllByTestedPipelineId?pipelineId={id}&page={pn}&size=1000").json()['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
503 ma_in_pip += pip
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
504
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
505 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
506 ma_in_pip = ma_in_pip['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
507
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
508 for g in ma_in_pip:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
509 d = {"Measurements": g[''], "Analysis": g['']}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
510 list_d.append(d)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
511
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
512 for i in list_d:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
513 mes.append(i[''])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
514 an.append(i[''])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
515
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
516 list_of_genes['Analysis'] = an
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
517 list_of_genes['Measurements'] = mes
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
518
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
519 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
520 list_of_genes.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
521 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
522 list_of_genes.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
523
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
524
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
525 # 19- Get all genes and measured values for a particular parameter
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
526 def par_gen(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
527 head = sys.argv[4]
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
528 id = inp
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
529
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
530 pa_query = f"{impc_api_search_url}/search/findAllByTestedParameterId?parameterId={id}&page=0&size=1000"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
531
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
532 gm_par = requests.get(pa_query).json()
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
533 pages = gm_par['page']['totalPages']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
534 max_elem = gm_par['page']['totalElements']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
535
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
536 print(f"Genes with {id}: {gm_par['page']['totalElements']}")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
537 d = {}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
538 list_d = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
539 list_of_genes = pd.DataFrame(columns=['Genes', 'Measured Values'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
540 gen = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
541 mes = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
542
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
543 if max_elem > 1000:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
544
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
545 gm_par = gm_par['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
546
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
547 for pn in range(1, pages):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
548 pip = requests.get(f"{impc_api_search_url}/search/findAllByTestedParameterId?parameterId={id}&page={pn}&size=1000").json()['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
549 gm_par += pip
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
550
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
551 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
552 gm_par = gm_par['_embedded']['genes']
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
553
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
554
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
555 for g in gm_par:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
556 d = {"Genes": g['mgiAccessionId'], "Measured Values": g['']}
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
557 list_d.append(d)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
558
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
559 for i in list_d:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
560 gen.append(i['Genes'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
561 mes.append(i['Measured Values'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
562
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
563 list_of_genes['Genes'] = gen
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
564 list_of_genes['Measured Values'] = mes
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
565
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
566 if head == 'True':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
567 list_of_genes.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
568 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
569 list_of_genes.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
570
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
571
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
572 def gene_mapping(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
573 tmp = inp.split(",")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
574 final_list = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
575 sym_list = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
576 for i in tmp:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
577 if 'MGI:' in i:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
578 final_list.append(i)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
579 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
580 sym_list.append(i)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
581 del (i)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
582 if len(sym_list) != 0:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
583 sym_list = ",".join(sym_list)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
584 biodbnet = f'https://biodbnet.abcc.ncifcrf.gov/webServices/rest.php/biodbnetRestApi.xml?method=db2db&format=row&input=genesymbol&inputValues={sym_list}&outputs=mgiid&taxonId=10090'
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
585 u = url.urlopen(biodbnet)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
586 db = pd.read_xml(u, elems_only=True)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
587 empty = True
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
588 discarded = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
589 for i in db.index:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
590 if db['MGIID'][i] != '-':
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
591 empty = False
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
592 final_list.append(db['MGIID'][i][4:])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
593 break
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
594 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
595 discarded.append(db['MGIID'][i][4:])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
596
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
597 if (len(db) == 0 and len(final_list) == 0) or (empty and len(final_list) == 0):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
598 stop_err("Error: it was not possible to map the input.")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
599 elif empty:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
600 print("Warning: it was not possible to map any of the gene symbols entry. Only MGI entries will be used.")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
601 elif len(discarded) != 0:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
602 print("Warning: it was not possible to map these elements: " + ",".join(discarded) + "\n")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
603 return(final_list)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
604
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
605 def pheno_mapping(inp):
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
606 tmp = inp.split(",")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
607 final_list = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
608 sym_list = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
609 for i in tmp:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
610 if 'MP:' in i:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
611 final_list.append(i)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
612 else:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
613 sym_list.append(i)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
614 del (i)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
615 if len(sym_list) != 0:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
616 url="https://raw.githubusercontent.com/AndreaFurlani/hp_mp_mapping_test/main/hp_mp_mapping.csv"
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
617 mapper = pd.read_csv(url,header=0,index_col=2)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
618 empty = True
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
619 discarded = []
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
620 for i in sym_list:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
621 try:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
622 final_list.append(mapper.loc[i]['mpId'])
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
623 empty=False
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
624 except KeyError:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
625 discarded.append(i)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
626 continue
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
627 if empty and len(final_list)==0:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
628 stop_err("Error: it was not possible to map the input.")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
629 elif empty:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
630 print("Warning: it was not possible to map any of the HP term entries. Only MP entries will be used.")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
631 elif len(discarded) != 0:
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
632 print("Warning: it was not possible to map these elements: " + ",".join(discarded) + "\n")
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
633 return (final_list)
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
634
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
635 if __name__ == "__main__":
0a9cf7f52b9c planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
infr
parents:
diff changeset
636 main()