htsworkflow/automation/solexa.py

   1 """Utilities to help process solexa/illumina runfolders
   2 """
   3 import os
   4 import re
   5
   6 def is_runfolder(name):
   7     """
   8     Is it a runfolder?
   9
  10     >>> print is_runfolder('090630_HWUSI-EAS999_0006_30LNFAAXX')
  11     True
  12     >>> print is_runfolder('hello')
  13     False
  14     """
  15     if re.match("^[0-9]{6}_[-A-Za-z0-9_]*$", name):
  16         return True
  17     else:
  18         return False
  19
  20 def get_top_dir(root, path):
  21     """
  22     Return the directory in path that is a subdirectory of root.
  23     e.g.
  24
  25     >>> print get_top_dir('/a/b/c', '/a/b/c/d/e/f')
  26     d
  27     >>> print get_top_dir('/a/b/c/', '/a/b/c/d/e/f')
  28     d
  29     >>> print get_top_dir('/a/b/c', '/g/e/f')
  30     None
  31     >>> print get_top_dir('/a/b/c', '/a/b/c')
  32     <BLANKLINE>
  33     """
  34     if path.startswith(root):
  35         subpath = path[len(root):]
  36         if subpath.startswith('/'):
  37             subpath = subpath[1:]
  38         return subpath.split(os.path.sep)[0]
  39     else:
  40         return None
  41