Mercurial > repos > melissacline > ucsc_xena_platform

annotate getXenaData.py @ 55:421b18a0b659 default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
update v17 step 2, add xena.jar
author jingchunzhu
date Tue, 22 Sep 2015 10:07:51 -0700
parents bb840cc2603d
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
52
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
1 #!/usr/bin/env python
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
2
41
02b0824c7d60 Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff changeset
3 import os, sys, string, json, csv
02b0824c7d60 Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff changeset
4 import xena_query as xena
02b0824c7d60 Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff changeset
5
52
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
6 if len(sys.argv[:])!=5:
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
7 print "python getXenaData.py hub datasetId outputfile galaxy_url\n"
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
8 sys.exit(2)
41
02b0824c7d60 Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff changeset
9
50
jingchunzhu
parents: 43
diff changeset
10 def main():
jingchunzhu
parents: 43
diff changeset
11 url = sys.argv[1]
jingchunzhu
parents: 43
diff changeset
12 dataset = sys.argv[2]
jingchunzhu
parents: 43
diff changeset
13 output = sys.argv[3]
52
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
14 GALAXY_URL = sys.argv[4]
41
02b0824c7d60 Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff changeset
15
50
jingchunzhu
parents: 43
diff changeset
16 fout = open(output,'w')
42
bc9784300015 more reliable, error handling
jingchunzhu <jingchunzhu@gmail.com>
parents: 41
diff changeset
17
50
jingchunzhu
parents: 43
diff changeset
18 contactUrl = url
jingchunzhu
parents: 43
diff changeset
19 if string.find(url,"galaxyxena") !=-1 and string.find(url,"ucsc.edu")!=-1:
jingchunzhu
parents: 43
diff changeset
20 contactUrl = "https://galaxyxena.soe.ucsc.edu:443/xena"
41
02b0824c7d60 Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff changeset
21
50
jingchunzhu
parents: 43
diff changeset
22 #testing if the url is reachable
jingchunzhu
parents: 43
diff changeset
23 try:
jingchunzhu
parents: 43
diff changeset
24 r =json.loads(xena.post(contactUrl, "(+ 1 2)"))
jingchunzhu
parents: 43
diff changeset
25
jingchunzhu
parents: 43
diff changeset
26 if r!=3.0:
52
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
27 fout.write("There is an error\n\n")
50
jingchunzhu
parents: 43
diff changeset
28 fout.write("You entered hub: %s\n" % (url))
52
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
29 fout.write("Possible causes for error:\n")
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
30 fout.write("1. The hub is not running.\n")
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
31 fout.write("2. The hub is not accessible to this galaxy instance at: %s.\n" %(GALAXY_URL))
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
32 fout.write("3. The hub url has a typo.\n")
50
jingchunzhu
parents: 43
diff changeset
33 fout.close()
jingchunzhu
parents: 43
diff changeset
34 sys.exit(1)
jingchunzhu
parents: 43
diff changeset
35
jingchunzhu
parents: 43
diff changeset
36 except:
52
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
37 fout.write("There is an error\n\n")
43
78d6e6772e30 more reliable
jingchunzhu <jingchunzhu@gmail.com>
parents: 42
diff changeset
38 fout.write("You entered hub: %s\n" % (url))
52
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
39 fout.write("Possible causes for error:\n")
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
40 fout.write("1. The hub is not running.\n")
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
41 fout.write("2. The hub is not accessible to this galaxy instance at: %s.\n" %(GALAXY_URL))
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
42 fout.write("3. The hub url has a typo.\n")
43
78d6e6772e30 more reliable
jingchunzhu <jingchunzhu@gmail.com>
parents: 42
diff changeset
43 fout.close()
78d6e6772e30 more reliable
jingchunzhu <jingchunzhu@gmail.com>
parents: 42
diff changeset
44 sys.exit(1)
42
bc9784300015 more reliable, error handling
jingchunzhu <jingchunzhu@gmail.com>
parents: 41
diff changeset
45
50
jingchunzhu
parents: 43
diff changeset
46 samples = xena.dataset_samples (contactUrl, dataset)
jingchunzhu
parents: 43
diff changeset
47 if not samples:
52
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
48 fout.write("There is an error\n\n")
50
jingchunzhu
parents: 43
diff changeset
49 fout.write("You entered dataset id: %s\n" % (dataset))
52
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
50 fout.write("Possible causes for error:\n")
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
51 fout.write("1. Dataset does not exist.\n")
50
jingchunzhu
parents: 43
diff changeset
52 fout.close()
jingchunzhu
parents: 43
diff changeset
53 sys.exit(1)
41
02b0824c7d60 Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff changeset
54
50
jingchunzhu
parents: 43
diff changeset
55 type = xena.dataset_type(contactUrl, dataset)
jingchunzhu
parents: 43
diff changeset
56 if type[0] not in ["genomicMatrix", "clinicalMatrix"]:
52
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
57 fout.write("There is an error\n\n")
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
58 fout.write("You entered dataset id: %s\n" % (dataset))
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
59 fout.write("It's datatype=%s\n" % (type[0]))
bb840cc2603d better error handling
jingchunzhu
parents: 50
diff changeset
60 fout.write("This type of data is not supported yet.\n")
50
jingchunzhu
parents: 43
diff changeset
61 fout.close()
jingchunzhu
parents: 43
diff changeset
62 sys.exit(1)
jingchunzhu
parents: 43
diff changeset
63
jingchunzhu
parents: 43
diff changeset
64 writer = csv.writer(fout, delimiter='\t')
jingchunzhu
parents: 43
diff changeset
65 writer.writerow(["sample"]+samples)
41
02b0824c7d60 Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff changeset
66
50
jingchunzhu
parents: 43
diff changeset
67 probes = xena.dataset_field(contactUrl, dataset)
jingchunzhu
parents: 43
diff changeset
68 start=0
jingchunzhu
parents: 43
diff changeset
69 size =100
jingchunzhu
parents: 43
diff changeset
70 N= len(probes)
jingchunzhu
parents: 43
diff changeset
71 for i in range (start, N,size):
jingchunzhu
parents: 43
diff changeset
72 results = xena.dataset_probe_values (contactUrl, dataset, samples, probes[i:i+size])
jingchunzhu
parents: 43
diff changeset
73 print ".",
jingchunzhu
parents: 43
diff changeset
74 for j in range (0, size):
jingchunzhu
parents: 43
diff changeset
75 if i+j == N:
jingchunzhu
parents: 43
diff changeset
76 break
jingchunzhu
parents: 43
diff changeset
77 writer.writerow([probes[i+j]]+results[j])
41
02b0824c7d60 Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff changeset
78
50
jingchunzhu
parents: 43
diff changeset
79 fout.close()
jingchunzhu
parents: 43
diff changeset
80 sys.exit(0)
41
02b0824c7d60 Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff changeset
81
50
jingchunzhu
parents: 43
diff changeset
82 if __name__ == "__main__":
jingchunzhu
parents: 43
diff changeset
83 main()
jingchunzhu
parents: 43
diff changeset
84