htsworkflow/pipelines/srf.py

   1 import logging
   2 import os
   3
   4 from htsworkflow.util import queuecommands
   5
   6 SOLEXA2SRF = 0
   7 ILLUMINA2SRF10 = 1
   8 ILLUMINA2SRF11 = 2
   9
  10 def pathname_to_run_name(base):
  11   """
  12   Convert a pathname to a base runfolder name
  13   handle the case with a trailing /
  14   """
  15   name = ""
  16   while len(name) == 0:
  17     base, name = os.path.split(base)
  18     if len(base) == 0:
  19       return None
  20   return name
  21
  22 def make_commands(run_name, lanes, site_name, destdir, cmdlevel=ILLUMINA2SRF11):
  23   """
  24   make a subprocess-friendly list of command line arguments to run solexa2srf
  25   generates files like:
  26   woldlab:080514_HWI-EAS229_0029_20768AAXX:8.srf
  27    site        run name                    lane
  28
  29   run_name - most of the file name (run folder name is a good choice)
  30   lanes - list of integers corresponding to which lanes to process
  31   site_name - name of your "sequencing site" or "Individual"
  32   destdir - where to write all the srf files
  33   """
  34   # clean up pathname
  35   logging.info("run_name %s" % ( run_name, ))
  36
  37   cmd_list = []
  38   for lane in lanes:
  39     name_prefix = '%s_%%l_%%t_' % (run_name,)
  40     destname = '%s_%s_%d.srf' % (site_name, run_name, lane)
  41     destdir = os.path.normpath(destdir)
  42     dest_path = os.path.join(destdir, destname)
  43     seq_pattern = 's_%d_*_seq.txt' % (lane,)
  44
  45     if cmdlevel == SOLEXA2SRF:
  46         cmd = ['solexa2srf',
  47                '-N', name_prefix,
  48                '-n', '%3x:%3y',
  49                '-o', dest_path,
  50                seq_pattern]
  51     elif cmdlevel == ILLUMINA2SRF10:
  52         cmd = ['illumina2srf',
  53                '-v1.0',
  54                '-o', dest_path,
  55                seq_pattern]
  56     elif cmdlevel == ILLUMINA2SRF11:
  57         seq_pattern = 's_%d_*_qseq.txt' % (lane,)
  58         cmd = ['illumina2srf',
  59                '-o', dest_path,
  60                seq_pattern]
  61     else:
  62         raise ValueError("Unrecognized run level %d" % (cmdlevel,))
  63
  64     logging.info("Generated command: " + " ".join(cmd))
  65     cmd_list.append(" ".join(cmd))
  66   return cmd_list
  67
  68 def run_srf_commands(bustard_dir, cmd_list, num_jobs):
  69     logging.info("chdir to %s" % (bustard_dir,))
  70     curdir = os.getcwd()
  71     os.chdir(bustard_dir)
  72     q = queuecommands.QueueCommands(cmd_list, num_jobs)
  73     q.run()
  74     os.chdir(curdir)
  75