Mercurial > repos > melissacline > ucsc_xena_platform
annotate getXenaData.py @ 50:3167c1a26101
fix
| author | jingchunzhu |
|---|---|
| date | Wed, 12 Aug 2015 16:38:12 -0700 |
| parents | 78d6e6772e30 |
| children | bb840cc2603d |
| rev | line source |
|---|---|
|
41
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
1 # getXenaData.py |
|
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
2 import os, sys, string, json, csv |
|
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
3 import xena_query as xena |
|
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
4 |
|
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
5 if len(sys.argv[:])!=4: |
|
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
6 print "python getXenaData.py hub datasetId outputfile\n" |
|
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
7 sys.exit(1) |
|
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
8 |
| 50 | 9 def main(): |
| 10 url = sys.argv[1] | |
| 11 dataset = sys.argv[2] | |
| 12 output = sys.argv[3] | |
|
41
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
13 |
| 50 | 14 fout = open(output,'w') |
|
42
bc9784300015
more reliable, error handling
jingchunzhu <jingchunzhu@gmail.com>
parents:
41
diff
changeset
|
15 |
| 50 | 16 contactUrl = url |
| 17 if string.find(url,"galaxyxena") !=-1 and string.find(url,"ucsc.edu")!=-1: | |
| 18 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
|
19 |
| 50 | 20 #testing if the url is reachable |
| 21 try: | |
| 22 r =json.loads(xena.post(contactUrl, "(+ 1 2)")) | |
| 23 | |
| 24 if r!=3.0: | |
| 25 print "The hub seems can not be reached, either it is not running, the url has a typo, or it is not accessible to this galaxy instance." | |
| 26 print "You entered hub: %s" % (url) | |
| 27 fout.write("The hub seems can not be reached, either it is not running, the url has a typo, or it is not accessible to this galaxy instance.\n") | |
| 28 fout.write("You entered hub: %s\n" % (url)) | |
| 29 fout.close() | |
| 30 sys.exit(1) | |
| 31 | |
| 32 except: | |
| 33 print "The hub seems can not be reached, either it is not running, the url has a typo, or it is not accessible to this galaxy instance." | |
| 43 | 34 print "You entered hub: %s" % (url) |
| 50 | 35 fout.write("The hub seems can not be reached, either it is not running, the url has a typo, or it is not accessible to this galaxy instance.\n") |
| 43 | 36 fout.write("You entered hub: %s\n" % (url)) |
| 37 fout.close() | |
| 38 sys.exit(1) | |
|
42
bc9784300015
more reliable, error handling
jingchunzhu <jingchunzhu@gmail.com>
parents:
41
diff
changeset
|
39 |
| 50 | 40 samples = xena.dataset_samples (contactUrl, dataset) |
| 41 if not samples: | |
| 42 print "Dataset does not exist" | |
| 43 print "You entered dataset id: %s" % (dataset) | |
| 44 fout.write("Dataset does not exists\n") | |
| 45 fout.write("You entered dataset id: %s\n" % (dataset)) | |
| 46 fout.close() | |
| 47 sys.exit(1) | |
|
41
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
48 |
| 50 | 49 type = xena.dataset_type(contactUrl, dataset) |
| 50 if type[0] not in ["genomicMatrix", "clinicalMatrix"]: | |
| 51 print "The type of data is not supported" | |
| 52 print "datatype=%s" % (type[0]) | |
| 53 fout.write("The type of data is not supported\n") | |
| 54 fout.write("datatype=%s\n" % (type[0])) | |
| 55 fout.close() | |
| 56 sys.exit(1) | |
| 57 | |
| 58 writer = csv.writer(fout, delimiter='\t') | |
| 59 writer.writerow(["sample"]+samples) | |
|
41
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
60 |
| 50 | 61 probes = xena.dataset_field(contactUrl, dataset) |
| 62 start=0 | |
| 63 size =100 | |
| 64 N= len(probes) | |
| 65 for i in range (start, N,size): | |
| 66 results = xena.dataset_probe_values (contactUrl, dataset, samples, probes[i:i+size]) | |
| 67 print ".", | |
| 68 for j in range (0, size): | |
| 69 if i+j == N: | |
| 70 break | |
| 71 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
|
72 |
| 50 | 73 fout.close() |
| 74 print "done" | |
| 75 sys.exit(0) | |
|
41
02b0824c7d60
Download data from any hub in the federated xena platform
jingchunzhu <jingchunzhu@gmail.com>
parents:
diff
changeset
|
76 |
| 50 | 77 |
| 78 if __name__ == "__main__": | |
| 79 main() | |
| 80 |