annotate impc_tool.py @ 0:4357848fb4e6 draft default tip

planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
author andrea.furlani
date Thu, 09 Jun 2022 10:51:23 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
1 import sys
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
2 import requests
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
3 import pandas as pd
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
4 import urllib.request as url
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
5
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
6 impc_api_url = "https://www.gentar.org/impc-dev-api/"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
7 impc_api_search_url = f"{impc_api_url}/genes"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
8 impc_api_gene_bundle_url = f"{impc_api_url}/geneBundles"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
9
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
10
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
11 def stop_err(msg):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
12 sys.exit(msg)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
13
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
14
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
15 def main():
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
16 inp = str(sys.argv[1])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
17 query = str(sys.argv[3])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
18
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
19 try:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
20 if query == '7':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
21 full_gene_table()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
22 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
23
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
24 if str(sys.argv[5])=="txt":
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
25 s = str(sys.argv[6])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
26 if s == "t":
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
27 sep = "\t"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
28 elif s == "s":
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
29 sep = " "
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
30 elif s in ",;.":
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
31 sep = s
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
32 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
33 sys.exit("Separator not valid, please change it.")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
34 inp = pd.read_csv(inp, header=None, delimiter=sep)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
35 if len(inp.columns)==1:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
36 inp = str(inp[0].values[0]).replace("'","")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
37 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
38 inp = inp.to_string(header=False, index=False).replace(" ",",")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
39
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
40 if query == '8':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
41 genes_in_pipeline(inp)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
42 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
43 elif query == '10': # it's here but not totally implemented
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
44 par_pip_ma(inp)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
45 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
46 elif query == '11': # it's here but not totally implemented
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
47 par_gen(inp)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
48 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
49 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
50 tmp = inp.split(",")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
51 final_list = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
52 sym_list = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
53 for i in tmp:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
54 if 'MGI:' in i or 'MP:' in i:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
55 final_list.append(i)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
56 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
57 sym_list.append(i)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
58 del(i)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
59 if len(sym_list) != 0:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
60 sym_list = ",".join(sym_list)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
61 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'
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
62 u = url.urlopen(biodbnet)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
63 db = pd.read_xml(u, elems_only=True)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
64 if len(db) == 0 and len(final_list) == 0:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
65 stop_err("It was not possible to map the input.")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
66 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
67 for i in range(0,len(db)):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
68 final_list.append(db['MGIID'][i][4:])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
69
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
70 inp= ",".join(final_list)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
71
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
72
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
73 if query == '1':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
74 get_pheno(inp)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
75 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
76 elif query == '2':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
77 get_genes(inp)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
78 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
79 elif query == '3':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
80 gene_set(inp)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
81 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
82 elif query == '4':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
83 extr_img(inp)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
84 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
85 elif query == '5':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
86 parameters(inp)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
87 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
88 elif query == '6':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
89 sign_par(inp)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
90 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
91 elif query == '9':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
92 sign_mp(inp)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
93 sys.exit(0)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
94 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
95 stop_err("Error, non-implemented query selected: " + query)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
96 except Exception as ex:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
97 stop_err('Error running get_pheno.py:\n' + str(ex))
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
98
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
99
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
100 # 1-Given a gene id, retrieve all the phenotypes related to it (id and name)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
101 def get_pheno(inp):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
102 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
103 mgi_accession_id = inp
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
104
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
105 gene_url = f"{impc_api_search_url}/{mgi_accession_id}"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
106 gene_data = requests.get(gene_url).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
107
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
108 p_list = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
109 id_list = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
110
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
111 if gene_data['significantMpTerms'] == None:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
112 stop_err("No significant MP terms found for this gene")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
113 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
114 for x in gene_data['significantMpTerms']:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
115 p_list.append(x['mpTermId'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
116 id_list.append(x['mpTermName'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
117
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
118 df = pd.DataFrame()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
119 df['MP term name'] = p_list
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
120 df['MP term id'] = id_list
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
121
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
122 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
123 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
124 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
125 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
126
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
127
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
128 # 3-Extract all genes having a particular phenotype or a set of phenotypes (e.g. relevant to a disease)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
129 def get_genes(inp):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
130 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
131 target_mp_terms = inp
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
132
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
133 ## All the data is paginated using the page and size parameters, by default the endpoint returns the first 20 hits
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
134 gene_by_phenotypes_query = f"{impc_api_search_url}/search/findAllBySignificantMpTermIdsContains?mpTermIds={target_mp_terms}&page=0&size=20"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
135 genes_with_clinical_chemistry_phenotypes = requests.get(gene_by_phenotypes_query).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
136 print(f"Genes with {target_mp_terms}: {genes_with_clinical_chemistry_phenotypes['page']['totalElements']}")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
137 list_of_genes = pd.DataFrame(columns=['Gene accession id', 'Gene name', 'Gene bundle url'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
138 acc = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
139 name = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
140 url = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
141
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
142 for gene in genes_with_clinical_chemistry_phenotypes['_embedded']['genes']:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
143 acc.append(gene['mgiAccessionId'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
144 name.append(gene['markerName'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
145 url.append(gene['_links']['geneBundle']['href'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
146
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
147 list_of_genes['Gene accession id'] = acc
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
148 list_of_genes['Gene name'] = name
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
149 list_of_genes['Gene bundle url'] = url
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
150
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
151 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
152 list_of_genes.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
153 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
154 list_of_genes.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
155
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
156 # 4. Extract all phenotypes which are present in a particular gene set (e.g. genes together in a pathway)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
157
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
158 def gene_set(inp):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
159 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
160 target_genes = inp
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
161
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
162 genes_in_gene_list_query = f"{impc_api_search_url}/search/findAllByMgiAccessionIdIn?mgiAccessionIds={target_genes}"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
163
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
164 genes_in_gene_list = requests.get(genes_in_gene_list_query).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
165 list_of_mp_terms_vs_gene_index = {}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
166
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
167 for gene in genes_in_gene_list['_embedded']['genes']:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
168 mp_terms = gene['significantMpTerms']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
169 gene_acc_id = gene["mgiAccessionId"]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
170 if mp_terms is None:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
171 continue
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
172 for mp_term_name in mp_terms:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
173 if mp_term_name['mpTermId'] not in list_of_mp_terms_vs_gene_index:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
174 list_of_mp_terms_vs_gene_index[mp_term_name['mpTermId']] = {"mp_term": mp_term_name['mpTermId'], "genes": []}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
175 list_of_mp_terms_vs_gene_index[mp_term_name['mpTermId']]["genes"].append(gene_acc_id)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
176 genes_by_mp_term = list(list_of_mp_terms_vs_gene_index.values())
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
177
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
178 df = pd.DataFrame()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
179 terms = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
180 genes = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
181 for i in genes_by_mp_term:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
182 terms.append(i['mp_term'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
183 genes.append(i['genes'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
184
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
185 df['mp_term'] = terms
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
186 df['genes'] = genes
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
187
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
188 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
189 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
190 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
191 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
192
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
193 # 7. Extract images with a particular phenotype or a set of phenotypes
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
194
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
195
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
196 def extr_img(inp):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
197 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
198 target_mp_terms = inp # ['MP:0002110', 'MP:0000559']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
199
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
200 ## All the data is paginated using the page and size parameters, by default the endpoint returns the first 20 hits
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
201 gene_by_phenotypes_query = f"{impc_api_search_url}/search/findAllBySignificantMpTermIdsContains?mpTermIds={target_mp_terms}&page=0&size=20"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
202 genes_with_morphology_mps = requests.get(gene_by_phenotypes_query).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
203 list_of_gene_bundle_urls = [gene["_links"]["geneBundle"]['href'] for gene in
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
204 genes_with_morphology_mps['_embedded']['genes']]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
205
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
206 gene_bundles = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
207 for gene_bundle_url in list_of_gene_bundle_urls:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
208 gene_bundle = requests.get(gene_bundle_url).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
209 gene_bundles.append(gene_bundle)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
210
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
211 images_with_morphology_mps = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
212
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
213 ## Doing just the first 20 and filtering out fields on the images
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
214 display_fields = ['geneSymbol', 'parameterName', 'biologicalSampleGroup', 'colonyId', 'zygosity', 'sex',
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
215 'downloadUrl', 'externalSampleId', 'thumbnailUrl']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
216
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
217
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
218 for gene_bundle in gene_bundles[:20]:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
219 if len(gene_bundle) == 4:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
220 continue
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
221 if gene_bundle["geneImages"] is not None:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
222 images = gene_bundle["geneImages"]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
223 for image in images:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
224 display_image = {k: v for k, v in image.items() if k in display_fields}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
225 images_with_morphology_mps.append(display_image)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
226
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
227 images_table = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
228 print(f"Images related to phenotype {target_mp_terms}: {len(images_with_morphology_mps)}")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
229 ## Displaying just the first 20 images
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
230 for i in images_with_morphology_mps[:20]:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
231 row = [f"<img src='{i['thumbnailUrl']}' />"] + list(i.values())
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
232 images_table.append(row)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
233
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
234 df = pd.DataFrame()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
235 externalSampleId = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
236 geneSymbol = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
237 biologicalSampleGroup = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
238 sex = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
239 colonyId = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
240 zygosity = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
241 parameterName = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
242 downloadUrl = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
243 thumbnailUrl = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
244
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
245 for i in images_table:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
246 externalSampleId.append(i[1])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
247 geneSymbol.append(i[2])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
248 biologicalSampleGroup.append(i[3])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
249 sex.append(i[4])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
250 colonyId.append(i[5])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
251 zygosity.append(i[6])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
252 parameterName.append(i[7])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
253 downloadUrl.append(i[8])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
254 thumbnailUrl.append(i[9])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
255
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
256 df['externalSampleId'] = externalSampleId
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
257 df['geneSymbol'] = geneSymbol
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
258 df['biologicalSampleGroup'] = biologicalSampleGroup
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
259 df['sex'] = sex
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
260 df['colonyId'] = colonyId
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
261 df['zygosity'] = zygosity
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
262 df['parameterName'] = parameterName
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
263 df['downloadUrl'] = downloadUrl
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
264 df['thumbnailUrl'] = thumbnailUrl
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
265
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
266 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
267 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
268 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
269 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
270
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
271 # 11- Which parameters have been measured for a particular knockout EASY
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
272
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
273
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
274 def parameters(inp):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
275 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
276 knockout = inp # "MGI:104636"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
277 gene_info = requests.get(impc_api_search_url + "/" + knockout).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
278
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
279 if gene_info['phenotypingDataAvailable']:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
280 geneBundle = requests.get(gene_info['_links']['geneBundle']['href']).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
281 gen_imgs = geneBundle['geneImages']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
282 par_list = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
283 l = {}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
284 for i in gen_imgs:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
285 l = {"Parameter Name": i['parameterName']}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
286 if l not in par_list:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
287 par_list.append(l)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
288 df = pd.DataFrame()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
289 l = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
290
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
291 for i in par_list:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
292 l.append(i['Parameter Name'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
293
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
294 df['Parameter'] = l
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
295 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
296 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
297 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
298 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
299
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
300 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
301 stop_err("No parameters available for this knockout gene")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
302
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
303
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
304 # 12- Which parameters identified a significant finding for a particular knockout line (colony) EASY
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
305 def sign_par(inp):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
306 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
307 knockout = inp # "MGI:104636"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
308
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
309 gene_info = requests.get(f"{impc_api_url}statisticalResults/search/findAllByMarkerAccessionIdIsAndSignificantTrue?mgiAccessionId=" + knockout).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
310 gene_stats = gene_info['_embedded']['statisticalResults']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
311
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
312 if len(gene_stats) == 0:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
313 stop_err("No statistically relevant parameters found for this knockout gene")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
314 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
315 df = pd.DataFrame()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
316 n = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
317 p = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
318 for g in gene_stats:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
319 n.append(g['parameterName'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
320 p.append(g['pvalue'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
321
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
322 df['Parameter name'] = n
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
323 df['p-value'] = p
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
324 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
325 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
326 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
327 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
328
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
329
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
330 # 13- List of genes names and ID measured in a pipeline
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
331 def genes_in_pipeline(inp):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
332 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
333 pip = inp
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
334
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
335 g_in_p_query = f"{impc_api_search_url}/search/findAllByTestedPipelineId?pipelineId={pip}&page=0&size=1000"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
336 genes_in_pip = requests.get(g_in_p_query).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
337 pages = genes_in_pip['page']['totalPages']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
338 max_elem = genes_in_pip['page']['totalElements']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
339
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
340 print(f"Genes with {pip}: {genes_in_pip['page']['totalElements']}")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
341 d ={ }
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
342 list_d = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
343 list_of_genes = pd.DataFrame(columns=['Gene accession id', 'Gene name'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
344 acc = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
345 name = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
346
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
347 if max_elem > 1000:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
348 g_in_p_query = genes_in_pip['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
349 for i in range(1,pages):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
350 gl = requests.get(f'{impc_api_search_url}/search/findAllByTestedPipelineId?pipelineId={pip}&page={i}&size=1000').json()['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
351 g_in_p_query += gl
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
352 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
353 g_in_p_query = genes_in_pip['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
354
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
355 for g in g_in_p_query:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
356 d = {"Gene Accession ID": g['mgiAccessionId'], "Gene Name": g['markerName']}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
357 list_d.append(d)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
358
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
359 for i in list_d:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
360 acc.append(i['Gene Accession ID'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
361 name.append(i['Gene Name'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
362
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
363 list_of_genes['Gene accession id'] = acc
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
364 list_of_genes['Gene name'] = name
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
365
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
366 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
367 list_of_genes.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
368 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
369 list_of_genes.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
370
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
371
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
372 # 14- Extract all genes and corresponding phenotypes related to a particular organ system(eg: significatMPTerm)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
373 def sign_mp(inp):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
374 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
375 mp_term = inp # ['MP:0005391']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
376
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
377 gene_by_mpterm_query = f"{impc_api_search_url}/search/findAllBySignificantMpTermIdsContains?mpTermIds={mp_term}&size=1000"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
378 genes_with_mpterm = requests.get(gene_by_mpterm_query).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
379
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
380 pages = genes_with_mpterm['page']['totalPages']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
381 genes_info = genes_with_mpterm['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
382
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
383 for pn in range(1,pages):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
384 pq = f"{impc_api_search_url}/search/findAllBySignificantMpTermIdsContains?mpTermIds={mp_term}&page={pn}&size=1000"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
385 g = requests.get(pq).json()['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
386 genes_info += g
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
387
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
388 list_d=[]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
389 d={}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
390 for g in genes_info:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
391 names=[]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
392 ids=[]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
393 for s in g['significantMpTerms']:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
394 names.append(s['mpTermName'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
395 ids.append(s['mpTermId'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
396 d={'Gene':g['mgiAccessionId'], 'mpTermId': ids, 'mpTermName':names}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
397 list_d.append(d)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
398
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
399
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
400 g = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
401 ids = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
402 names = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
403 for i in list_d:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
404 g.append(i['Gene'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
405 ids.append(i['mpTermId'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
406 names.append(i['mpTermName'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
407
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
408 df = pd.DataFrame()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
409 df['Gene Id']=g
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
410 df['Significant MP terms Ids']=ids
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
411 df['Significant MP terms Names']=names
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
412
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
413 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
414 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
415 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
416 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
417
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
418
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
419 # 16- Full table of genes and all identified phenotypes
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
420
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
421 def full_gene_table():
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
422 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
423 gene_list = requests.get(impc_api_search_url + '?page=0&size=1000').json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
424 pages = gene_list['page']['totalPages']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
425 genes_info = gene_list['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
426
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
427 for pn in range(1,pages):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
428 gp = requests.get(impc_api_search_url + f'?page={pn}&size=1000').json()['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
429 genes_info += gp
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
430
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
431 d = {}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
432 list_d=[]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
433
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
434 for i in genes_info:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
435 l = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
436 if i['significantMpTerms'] is None:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
437 d={"Gene": i['mgiAccessionId'], "Identified phenotypes": "None"}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
438 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
439 d = {"Gene": i['mgiAccessionId'], "Identified phenotypes": [sub['mpTermId'] for sub in i['significantMpTerms']]}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
440 list_d.append(d)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
441
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
442 df = pd.DataFrame()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
443 g = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
444 p = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
445 for i in list_d:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
446 g.append(i['Gene'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
447 p.append(i['Identified phenotypes'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
448
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
449 df['MGI id'] = g
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
450 df['MP term list'] = p
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
451
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
452 for i in range(0, len(df)):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
453 if df['MP term list'][i] != "None":
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
454 df['MP term list'][i] = str(df['MP term list'][i])[1:-1].replace("'", "")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
455
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
456 if str(sys.argv[1]) == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
457 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
458 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
459 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
460 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
461 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
462 df = df[df['MP term list'] != "None"]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
463 df.reset_index(drop=True, inplace=True)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
464 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
465 df.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
466 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
467 df.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
468
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
469 # Old method, chech which is faster
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
470 # max_elem = gene_list['page']['totalElements']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
471 # d = {}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
472 # list_d = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
473 # for i in range(0, pages):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
474 # gl = requests.get(impc_api_search_url + '?page=' + str(i) + '&size=' + str(max_elem)).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
475 # for g in gl['_embedded']['genes']:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
476 # if g['significantMpTerms'] is None:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
477 # d = {"Gene": g['mgiAccessionId'], "Identified phenotypes": "None"}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
478 # else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
479 # d = {"Gene": g['mgiAccessionId'], "Identified phenotypes": [ sub['mpTermId'] for sub in g['significantMpTerms'] ]}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
480 # list_d.append(d)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
481
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
482
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
483
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
484
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
485 # 18- Extract measurements and analysis for a parameter or pipeline
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
486
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
487 def par_pip_ma(inp):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
488 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
489 id = inp
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
490
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
491 if id[0:4] == "IMPC":
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
492 par = True
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
493 ma_query = f"{impc_api_search_url}/search/findAllByTestedParameterId?parameterId={id}&page=0&size=1000"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
494 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
495 ma_query = f"{impc_api_search_url}/search/findAllByTestedPipelineId?pipelineId={id}&page=0&size=1000"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
496 par = False
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
497
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
498 ma_in_pip = requests.get(ma_query).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
499 pages = ma_in_pip['page']['totalPages']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
500 max_elem = ma_in_pip['page']['totalElements']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
501
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
502 print(f"Genes with {id}: {ma_in_pip['page']['totalElements']}")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
503 d = {}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
504 list_d = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
505 list_of_genes = pd.DataFrame(columns=['Measurements', 'Analysis'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
506 mes = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
507 an = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
508
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
509 if max_elem > 1000:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
510
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
511 ma_in_pip = ma_in_pip['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
512 for pn in range(1, pages):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
513 if par:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
514 pip = requests.get(f"{impc_api_search_url}/search/findAllByTestedParameterId?parameterId={id}&page={pn}&size=1000").json()['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
515 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
516 pip = requests.get(f"{impc_api_search_url}/search/findAllByTestedPipelineId?pipelineId={id}&page={pn}&size=1000").json()['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
517 ma_in_pip += pip
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
518
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
519 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
520 ma_in_pip = ma_in_pip['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
521
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
522 for g in ma_in_pip:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
523 d = {"Measurements": g[''], "Analysis": g['']}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
524 list_d.append(d)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
525
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
526 for i in list_d:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
527 mes.append(i[''])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
528 an.append(i[''])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
529
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
530 list_of_genes['Analysis'] = an
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
531 list_of_genes['Measurements'] = mes
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
532
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
533 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
534 list_of_genes.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
535 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
536 list_of_genes.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
537
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
538
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
539 # 19- Get all genes and measured values for a particular parameter
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
540 def par_gen(inp):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
541 head = sys.argv[4]
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
542 id = inp
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
543
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
544 pa_query = f"{impc_api_search_url}/search/findAllByTestedParameterId?parameterId={id}&page=0&size=1000"
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
545
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
546 gm_par = requests.get(pa_query).json()
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
547 pages = gm_par['page']['totalPages']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
548 max_elem = gm_par['page']['totalElements']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
549
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
550 print(f"Genes with {id}: {gm_par['page']['totalElements']}")
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
551 d = {}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
552 list_d = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
553 list_of_genes = pd.DataFrame(columns=['Genes', 'Measured Values'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
554 gen = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
555 mes = []
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
556
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
557 if max_elem > 1000:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
558
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
559 gm_par = gm_par['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
560
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
561 for pn in range(1, pages):
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
562 pip = requests.get(f"{impc_api_search_url}/search/findAllByTestedParameterId?parameterId={id}&page={pn}&size=1000").json()['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
563 gm_par += pip
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
564
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
565 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
566 gm_par = gm_par['_embedded']['genes']
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
567
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
568
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
569 for g in gm_par:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
570 d = {"Genes": g['mgiAccessionId'], "Measured Values": g['']}
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
571 list_d.append(d)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
572
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
573 for i in list_d:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
574 gen.append(i['Genes'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
575 mes.append(i['Measured Values'])
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
576
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
577 list_of_genes['Genes'] = gen
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
578 list_of_genes['Measured Values'] = mes
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
579
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
580 if head == 'True':
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
581 list_of_genes.to_csv(sys.argv[2], header=True, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
582 else:
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
583 list_of_genes.to_csv(sys.argv[2], header=False, index=False, sep="\t", index_label=False)
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
584
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
585
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
586 if __name__ == "__main__":
4357848fb4e6 planemo upload commit 213f6eeb03f96bb13d0ace6e0c87e2562d37f728-dirty
andrea.furlani
parents:
diff changeset
587 main()