htsworkflow/submission/ucsc.py

   1 """Utilities for extracting information from the ENCODE DCC
   2 """
   3 import urlparse
   4 import urllib2
   5
   6 UCSCEncodePipeline = "http://encodesubmit.ucsc.edu/pipeline/"
   7
   8
   9 def ddf_download_url(submission_id):
  10     """Return url to download a DDF for a submission
  11
  12     >>> ddf_download_url(1234)
  13     'http://encodesubmit.ucsc.edu/pipeline/download_ddf/1234'
  14     """
  15     fragment = 'download_ddf/%s' % (submission_id,)
  16     return urlparse.urljoin(UCSCEncodePipeline, fragment)
  17
  18
  19 def daf_download_url(submission_id):
  20     """Return url to download a DAF for a submission
  21
  22     >>> daf_download_url(1234)
  23     'http://encodesubmit.ucsc.edu/pipeline/download_daf/1234'
  24     """
  25     fragment = 'download_daf/%s' % (submission_id,)
  26     return urlparse.urljoin(UCSCEncodePipeline, fragment)
  27
  28
  29 def submission_view_url(submission_id):
  30     """Return url to download a DAF for a submission
  31
  32     >>> submission_view_url(1234)
  33     'http://encodesubmit.ucsc.edu/pipeline/show/1234'
  34     """
  35     fragment = 'show/%s' % (submission_id,)
  36     return urlparse.urljoin(UCSCEncodePipeline, fragment)
  37
  38
  39 def get_ucsc_file_index(base_url):
  40     """Get index of files for a ENCODE collection
  41     """
  42     if base_url[-1] != '/': base_url += '/'
  43     request = urllib2.urlopen(base_url + 'files.txt')
  44     file_index = parse_ucsc_file_index(request)
  45     return file_index
  46
  47
  48 def parse_ucsc_file_index(stream):
  49     """Turn a UCSC DCC files.txt index into a dictionary of name-value pairs
  50     """
  51     file_index = {}
  52     for line in stream:
  53         filename, attribute_line = line.split('\t')
  54         attributes = {}
  55         for assignment in  attribute_line.split(';'):
  56             name, value = assignment.split('=')
  57             attributes[name.strip()] = value.strip()
  58
  59         file_index[filename] = attributes
  60     return file_index