Bed file generation is here!

[htsworkflow.git] / gaworkflow / util / makebed.py

diff --git a/gaworkflow/util/makebed.py b/gaworkflow/util/makebed.py
index 738b815b270d2d8352d6157a7b5e1948b437e3a3..4d73cb6cbcad46b73481529a311ae86dd926ec3d 100755 (executable)
--- a/gaworkflow/util/makebed.py
+++ b/gaworkflow/util/makebed.py
@@ -41,6 +41,45 @@ def make_bed_from_eland_stream(instream, outstream, name, description, chromosom
       sense_color[fields[SENSE]],
       os.linesep  
     ))
+    
+def make_bed_from_eland_stream_generator(instream, name, description, chromosome_prefix='chr'):
+  """
+  read an eland result file from instream and output it as a generator (iterator)
+  """
+  # indexes into fields in eland_result.txt file
+  SEQ = 1
+  CHR = 6
+  START = 7
+  SENSE = 8
+  # map eland_result.txt sense 
+  sense_map = { 'F': '+', 'R': '-'}
+  sense_color = { 'F': '0,0,255', 'R': '255,255,0' }
+  # provide default track names
+  if name is None: name = "track"
+  if description is None: description = "eland result file"
+  bed_header = 'track name="%s" description="%s" visibility=4 itemRgb="ON"'
+  bed_header += os.linesep
+  yield bed_header % (name, description)
+
+  for line in instream:
+    fields = line.split()
+    # we need more than the CHR field, and it needs to match a chromosome
+    if len(fields) <= CHR or \
+          (chromosome_prefix is not None and \
+             fields[CHR][:3] != chromosome_prefix):
+      continue
+    start = fields[START]
+    stop = int(start) + len(fields[SEQ])
+    chromosome, extension = fields[CHR].split('.')
+    assert extension == "fa"
+    yield '%s %s %d read 0 %s - - %s%s' % (
+      chromosome,
+      start,
+      stop,
+      sense_map[fields[SENSE]], 
+      sense_color[fields[SENSE]],
+      os.linesep  
+    )
 
 def make_description(database, flowcell_id, lane):
     """
@@ -49,7 +88,7 @@ def make_description(database, flowcell_id, lane):
     from gaworkflow.util.fctracker import fctracker
 
     fc = fctracker(database)
-    cells = fc._get_flowcells("where flowcell_id='%s'" % (flowcell_id))
+    cells = fc._get_flowcells("where flowcell_id LIKE '%s%%'" % (flowcell_id))
     if len(cells) != 1:
       raise RuntimeError("couldn't find flowcell id %s" % (flowcell_id))
     lane = int(lane)
@@ -59,7 +98,11 @@ def make_description(database, flowcell_id, lane):
     name = "%s-%s" % (flowcell_id, lane)
 
     cell_id, cell = cells.items()[0]
-    assert cell_id == flowcell_id
+    
+    #The assertion is no longer true after I changed
+    # the where statement to include the LIKE command.
+    # because flowcells are being renamed to 'FC12269 (deleted)'
+    #assert cell_id == flowcell_id
 
     cell_library_id = cell['lane_%d_library_id' %(lane,)]
     cell_library = cell['lane_%d_library' %(lane,)]