"""
Extract configuration from Illumina Bustard Directory.
-This includes the version number, run date, bustard executable parameters, and
-phasing estimates.
+This includes the version number, run date, bustard executable parameters, and
+phasing estimates.
"""
from copy import copy
from datetime import date
import sys
import time
-from htsworkflow.pipelines.runfolder import \
+from htsworkflow.pipelines import \
ElementTree, \
VERSION_RE, \
EUROPEAN_STRPTIME
+LOGGER = logging.getLogger(__name__)
+
# make epydoc happy
__docformat__ = "restructuredtext en"
self.base['A'] = [ float(v) for v in data[:4] ]
self.base['C'] = [ float(v) for v in data[4:8] ]
self.base['G'] = [ float(v) for v in data[8:12] ]
- self.base['T'] = [ float(v) for v in data[12:16] ]
+ self.base['T'] = [ float(v) for v in data[12:16] ]
else:
raise RuntimeError("matrix file %s is unusual" % (pathname,))
def get_elements(self):
# we estimated the matrix from something in this run.
# though we don't really care which lane it was
if matrix_auto_lane == 0:
- auto_lane_fragment = ""
+ # its defaulting to all of the lanes, so just pick one
+ auto_lane_fragment = "_1"
else:
auto_lane_fragment = "_%d" % ( matrix_auto_lane,)
-
+
for matrix_name in ['s%s_02_matrix.txt' % (auto_lane_fragment,),
's%s_1_matrix.txt' % (auto_lane_fragment,),
]:
BUSTARD_CONFIG = 'BaseCallAnalysis'
def __init__(self, xml=None):
- self.version = None
+ self._path_version = None # version number from directory name
self.date = None
self.user = None
self.phasing = {}
if xml is not None:
self.set_elements(xml)
+ def update_attributes_from_pathname(self):
+ """Update version, date, user from bustard directory names
+ Obviously this wont work for BaseCalls or later runfolders
+ """
+ if self.pathname is None:
+ raise ValueError(
+ "Set pathname before calling update_attributes_from_pathname")
+ path, name = os.path.split(self.pathname)
+
+ if not re.match('bustard', name, re.IGNORECASE):
+ return
+
+ groups = name.split("_")
+ version = re.search(VERSION_RE, groups[0])
+ self._path_version = version.group(1)
+ t = time.strptime(groups[1], EUROPEAN_STRPTIME)
+ self.date = date(*t[0:3])
+ self.user = groups[2]
+
+ def _get_sequence_format(self):
+ """Guess sequence format"""
+ project_glob = os.path.join(self.pathname, 'Project_*')
+ LOGGER.debug("Scanning: %s" % (project_glob,))
+ projects = glob(project_glob)
+ if len(projects) > 0:
+ # Hey we look like a demultiplexed run
+ return 'fastq'
+ seqs = glob(os.path.join(self.pathname, '*_seq.txt'))
+ if len(seqs) > 0:
+ return 'srf'
+ return 'qseq'
+ sequence_format = property(_get_sequence_format)
+
+ def _get_software_version(self):
+ """return software name, version tuple"""
+ if self.bustard_config is None:
+ if self._path_version is not None:
+ return 'Bustard', self._path_version
+ else:
+ return None
+ software_nodes = self.bustard_config.xpath('Run/Software')
+ if len(software_nodes) == 0:
+ return None
+ elif len(software_nodes) > 1:
+ raise RuntimeError("Too many software XML elements for bustard.py")
+ else:
+ return (software_nodes[0].attrib['Name'],
+ software_nodes[0].attrib['Version'])
+
+ def _get_software(self):
+ """Return software name"""
+ software_version = self._get_software_version()
+ return software_version[0] if software_version is not None else None
+ software = property(_get_software)
+
+ def _get_version(self):
+ """Return software name"""
+ software_version = self._get_software_version()
+ return software_version[1] if software_version is not None else None
+ version = property(_get_version)
+
+
def _get_time(self):
if self.date is None:
return None
ElementTree.dump(self.get_elements())
def get_elements(self):
- root = ElementTree.Element('Bustard',
+ root = ElementTree.Element('Bustard',
{'version': str(Bustard.XML_VERSION)})
version = ElementTree.SubElement(root, Bustard.SOFTWARE_VERSION)
version.text = self.version
# add crosstalk matrix if it exists
if self.crosstalk is not None:
root.append(self.crosstalk.get_elements())
-
+
# add bustard config if it exists
if self.bustard_config is not None:
root.append(self.bustard_config)
raise ValueError('Expected "Bustard" SubElements')
xml_version = int(tree.attrib.get('version', 0))
if xml_version > Bustard.XML_VERSION:
- logging.warn('Bustard XML tree is a higher version than this class')
+ LOGGER.warn('Bustard XML tree is a higher version than this class')
for element in list(tree):
if element.tag == Bustard.SOFTWARE_VERSION:
- self.version = element.text
+ self._path_version = element.text
elif element.tag == Bustard.DATE:
self.date = date.fromtimestamp(float(element.text))
elif element.tag == Bustard.USER:
self.bustard_config = element
else:
raise ValueError("Unrecognized tag: %s" % (element.tag,))
-
+
def bustard(pathname):
"""
Construct a Bustard object by analyzing an Illumina Bustard directory.
:Return:
Fully initialized Bustard object.
"""
- b = Bustard()
pathname = os.path.abspath(pathname)
+ bustard_filename = os.path.join(pathname, 'config.xml')
+ demultiplexed_filename = os.path.join(pathname,
+ 'DemultiplexedBustardConfig.xml')
+
+ if os.path.exists(demultiplexed_filename):
+ b = bustard_from_hiseq(pathname, demultiplexed_filename)
+ elif os.path.exists(bustard_filename):
+ b = bustard_from_ga2(pathname, bustard_filename)
+ else:
+ b = bustard_from_ga1(pathname)
+
+ return b
+
+def bustard_from_ga1(pathname):
+ """Initialize bustard class from ga1 era runfolders.
+ """
path, name = os.path.split(pathname)
+
groups = name.split("_")
- if groups[0].lower().startswith('bustard'):
- version = re.search(VERSION_RE, groups[0])
- b.version = version.group(1)
- t = time.strptime(groups[1], EUROPEAN_STRPTIME)
- b.date = date(*t[0:3])
- b.user = groups[2]
- elif groups[0] == 'BaseCalls':
- # stub values
- b.version = None
- b.date = None
- b.user = None
+ if len(groups) < 3:
+ msg = "Not enough information to create attributes"\
+ " from directory name: %s"
+ LOGGER.error(msg % (pathname,))
+ return None
+ b = Bustard()
b.pathname = pathname
- bustard_config_filename = os.path.join(pathname, 'config.xml')
- paramfiles = glob(os.path.join(pathname, "params?.xml"))
- for paramfile in paramfiles:
- phasing = Phasing(paramfile)
- assert (phasing.lane >= 1 and phasing.lane <= 8)
- b.phasing[phasing.lane] = phasing
+ b.update_attributes_from_pathname()
+ version = re.search(VERSION_RE, groups[0])
+ b._path_version = version.group(1)
+ t = time.strptime(groups[1], EUROPEAN_STRPTIME)
+ b.date = date(*t[0:3])
+ b.user = groups[2]
+
# I only found these in Bustard1.9.5/1.9.6 directories
if b.version in ('1.9.5', '1.9.6'):
# at least for our runfolders for 1.9.5 and 1.9.6 matrix[1-8].txt are always the same
crosstalk_file = os.path.join(pathname, "matrix1.txt")
b.crosstalk = CrosstalkMatrix(crosstalk_file)
+
+ add_phasing(b)
+ return b
+
+
+def bustard_from_ga2(pathname, config_filename):
+ """Initialize bustard class from ga2-era runfolder
+ Actually I don't quite remember if it is exactly the GA2s, but
+ its after the original runfolder style and before the HiSeq.
+ """
# for version 1.3.2 of the pipeline the bustard version number went down
# to match the rest of the pipeline. However there's now a nifty
# new (useful) bustard config file.
- elif os.path.exists(bustard_config_filename):
- bustard_config_root = ElementTree.parse(bustard_config_filename)
- b.bustard_config = bustard_config_root.getroot()
- b.crosstalk = crosstalk_matrix_from_bustard_config(b.pathname, b.bustard_config)
- software = bustard_config_root.find('*/Software')
- b.version = software.attrib['Version']
- #b.version = software.attrib['Name'] + "-" + software.attrib['Version']
+
+ # stub values
+ b = Bustard()
+ b.pathname = pathname
+ b.update_attributes_from_pathname()
+ bustard_config_root = ElementTree.parse(config_filename)
+ b.bustard_config = bustard_config_root.getroot()
+ b.crosstalk = crosstalk_matrix_from_bustard_config(b.pathname,
+ b.bustard_config)
+ add_phasing(b)
return b
+def bustard_from_hiseq(pathname, config_filename):
+ b = Bustard()
+ b.pathname = pathname
+ bustard_config_root = ElementTree.parse(config_filename)
+ b.bustard_config = bustard_config_root.getroot()
+ add_phasing(b)
+ return b
+
+def add_phasing(bustard_obj):
+ paramfiles = glob(os.path.join(bustard_obj.pathname,
+ "params?.xml"))
+ for paramfile in paramfiles:
+ phasing = Phasing(paramfile)
+ assert (phasing.lane >= 1 and phasing.lane <= 8)
+ bustard_obj.phasing[phasing.lane] = phasing
+
def fromxml(tree):
"""
Reconstruct a htsworkflow.pipelines.Bustard object from an xml block
for bustard_dir in args:
print u'analyzing bustard directory: ' + unicode(bustard_dir)
bustard_object = bustard(bustard_dir)
- bustard_object.dump()
+ bustard_object.dump()
bustard_object2 = Bustard(xml=bustard_object.get_elements())
print ('-------------------------------------')