gaworkflow/pipeline/gerald.py

   1 """
   2 Provide access to information stored in the GERALD directory.
   3 """
   4 from datetime import datetime, date
   5 from glob import glob
   6 import logging
   7 import os
   8 import stat
   9 import time
  10
  11 from gaworkflow.pipeline.runfolder import \
  12    ElementTree, \
  13    EUROPEAN_STRPTIME, \
  14    LANES_PER_FLOWCELL, \
  15    VERSION_RE
  16 from gaworkflow.util.ethelp import indent, flatten
  17 from gaworkflow.util.opener import autoopen
  18
  19 class Gerald(object):
  20     """
  21     Capture meaning out of the GERALD directory
  22     """
  23     XML_VERSION = 1
  24     GERALD='Gerald'
  25     RUN_PARAMETERS='RunParameters'
  26     SUMMARY='Summary'
  27
  28     class LaneParameters(object):
  29         """
  30         Make it easy to access elements of LaneSpecificRunParameters from python
  31         """
  32         def __init__(self, gerald, key):
  33             self._gerald = gerald
  34             self._key = key
  35
  36         def __get_attribute(self, xml_tag):
  37             subtree = self._gerald.tree.find('LaneSpecificRunParameters')
  38             container = subtree.find(xml_tag)
  39             if container is None:
  40                 return None
  41             if len(container.getchildren()) != LANES_PER_FLOWCELL:
  42                 raise RuntimeError('GERALD config.xml file changed')
  43             lanes = [x.tag.split('_')[1] for x in container.getchildren()]
  44             index = lanes.index(self._key)
  45             element = container[index]
  46             return element.text
  47         def _get_analysis(self):
  48             return self.__get_attribute('ANALYSIS')
  49         analysis = property(_get_analysis)
  50
  51         def _get_eland_genome(self):
  52             genome = self.__get_attribute('ELAND_GENOME')
  53             # default to the chipwide parameters if there isn't an
  54             # entry in the lane specific paramaters
  55             if genome is None:
  56                 subtree = self._gerald.tree.find('ChipWideRunParameters')
  57                 container = subtree.find('ELAND_GENOME')
  58                 genome = container.text
  59             return genome
  60         eland_genome = property(_get_eland_genome)
  61
  62         def _get_read_length(self):
  63             return self.__get_attribute('READ_LENGTH')
  64         read_length = property(_get_read_length)
  65
  66         def _get_use_bases(self):
  67             return self.__get_attribute('USE_BASES')
  68         use_bases = property(_get_use_bases)
  69
  70     class LaneSpecificRunParameters(object):
  71         """
  72         Provide access to LaneSpecificRunParameters
  73         """
  74         def __init__(self, gerald):
  75             self._gerald = gerald
  76             self._keys = None
  77         def __getitem__(self, key):
  78             return Gerald.LaneParameters(self._gerald, key)
  79         def keys(self):
  80             if self._keys is None:
  81                 tree = self._gerald.tree
  82                 analysis = tree.find('LaneSpecificRunParameters/ANALYSIS')
  83                 # according to the pipeline specs I think their fields
  84                 # are sampleName_laneID, with sampleName defaulting to s
  85                 # since laneIDs are constant lets just try using
  86                 # those consistently.
  87                 self._keys = [ x.tag.split('_')[1] for x in analysis]
  88             return self._keys
  89         def values(self):
  90             return [ self[x] for x in self.keys() ]
  91         def items(self):
  92             return zip(self.keys(), self.values())
  93         def __len__(self):
  94             return len(self.keys())
  95
  96     def __init__(self, xml=None):
  97         self.pathname = None
  98         self.tree = None
  99
 100         # parse lane parameters out of the config.xml file
 101         self.lanes = Gerald.LaneSpecificRunParameters(self)
 102
 103         self.summary = None
 104         self.eland_results = None
 105
 106         if xml is not None:
 107             self.set_elements(xml)
 108
 109     def _get_date(self):
 110         if self.tree is None:
 111             return datetime.today()
 112         timestamp = self.tree.findtext('ChipWideRunParameters/TIME_STAMP')
 113         epochstamp = time.mktime(time.strptime(timestamp, '%c'))
 114         return datetime.fromtimestamp(epochstamp)
 115     date = property(_get_date)
 116
 117     def _get_time(self):
 118         return time.mktime(self.date.timetuple())
 119     time = property(_get_time, doc='return run time as seconds since epoch')
 120
 121     def _get_version(self):
 122         if self.tree is None:
 123             return None
 124         return self.tree.findtext('ChipWideRunParameters/SOFTWARE_VERSION')
 125     version = property(_get_version)
 126
 127     def dump(self):
 128         """
 129         Debugging function, report current object
 130         """
 131         print 'Gerald version:', self.version
 132         print 'Gerald run date:', self.date
 133         print 'Gerald config.xml:', self.tree
 134         self.summary.dump()
 135
 136     def get_elements(self):
 137         if self.tree is None or self.summary is None:
 138             return None
 139
 140         gerald = ElementTree.Element(Gerald.GERALD,
 141                                      {'version': unicode(Gerald.XML_VERSION)})
 142         gerald.append(self.tree)
 143         gerald.append(self.summary.get_elements())
 144         if self.eland_results:
 145             gerald.append(self.eland_results.get_elements())
 146         return gerald
 147
 148     def set_elements(self, tree):
 149         if tree.tag !=  Gerald.GERALD:
 150             raise ValueError('exptected GERALD')
 151         xml_version = int(tree.attrib.get('version', 0))
 152         if xml_version > Gerald.XML_VERSION:
 153             logging.warn('XML tree is a higher version than this class')
 154         for element in list(tree):
 155             tag = element.tag.lower()
 156             if tag == Gerald.RUN_PARAMETERS.lower():
 157                 self.tree = element
 158             elif tag == Gerald.SUMMARY.lower():
 159                 self.summary = Summary(xml=element)
 160             elif tag == ELAND.ELAND.lower():
 161                 self.eland_results = ELAND(xml=element)
 162             else:
 163                 logging.warn("Unrecognized tag %s" % (element.tag,))
 164
 165
 166 def gerald(pathname):
 167     g = Gerald()
 168     g.pathname = pathname
 169     path, name = os.path.split(pathname)
 170     config_pathname = os.path.join(pathname, 'config.xml')
 171     g.tree = ElementTree.parse(config_pathname).getroot()
 172
 173     # parse Summary.htm file
 174     summary_pathname = os.path.join(pathname, 'Summary.htm')
 175     g.summary = Summary(summary_pathname)
 176     # parse eland files
 177     g.eland_results = eland(g.pathname, g)
 178     return g
 179
 180 def tonumber(v):
 181     """
 182     Convert a value to int if its an int otherwise a float.
 183     """
 184     try:
 185         v = int(v)
 186     except ValueError, e:
 187         v = float(v)
 188     return v
 189
 190 def parse_mean_range(value):
 191     """
 192     Parse values like 123 +/- 4.5
 193     """
 194     if value.strip() == 'unknown':
 195         return 0, 0
 196
 197     average, pm, deviation = value.split()
 198     if pm != '+/-':
 199         raise RuntimeError("Summary.htm file format changed")
 200     return tonumber(average), tonumber(deviation)
 201
 202 def make_mean_range_element(parent, name, mean, deviation):
 203     """
 204     Make an ElementTree subelement <Name mean='mean', deviation='deviation'/>
 205     """
 206     element = ElementTree.SubElement(parent, name,
 207                                      { 'mean': unicode(mean),
 208                                        'deviation': unicode(deviation)})
 209     return element
 210
 211 def parse_mean_range_element(element):
 212     """
 213     Grab mean/deviation out of element
 214     """
 215     return (tonumber(element.attrib['mean']),
 216             tonumber(element.attrib['deviation']))
 217
 218 class Summary(object):
 219     """
 220     Extract some useful information from the Summary.htm file
 221     """
 222     XML_VERSION = 1
 223     SUMMARY = 'Summary'
 224
 225     class LaneResultSummary(object):
 226         """
 227         Parse the LaneResultSummary table out of Summary.htm
 228         Mostly for the cluster number
 229         """
 230         LANE_RESULT_SUMMARY = 'LaneResultSummary'
 231         TAGS = {
 232           'Cluster': 'cluster',
 233           'AverageFirstCycleIntensity': 'average_first_cycle_intensity',
 234           'PercentIntensityAfter20Cycles': 'percent_intensity_after_20_cycles',
 235           'PercentPassFilterClusters': 'percent_pass_filter_clusters',
 236           'PercentPassFilterAlign': 'percent_pass_filter_align',
 237           'AverageAlignmentScore': 'average_alignment_score',
 238           'PercentErrorRate': 'percent_error_rate'
 239         }
 240
 241         def __init__(self, html=None, xml=None):
 242             self.lane = None
 243             self.cluster = None
 244             self.average_first_cycle_intensity = None
 245             self.percent_intensity_after_20_cycles = None
 246             self.percent_pass_filter_clusters = None
 247             self.percent_pass_filter_align = None
 248             self.average_alignment_score = None
 249             self.percent_error_rate = None
 250
 251             if html is not None:
 252                 self.set_elements_from_html(html)
 253             if xml is not None:
 254                 self.set_elements(xml)
 255
 256         def set_elements_from_html(self, row_element):
 257             data = [ flatten(x) for x in row_element ]
 258             if len(data) != 8:
 259                 raise RuntimeError("Summary.htm file format changed")
 260
 261             self.lane = data[0]
 262             self.cluster = parse_mean_range(data[1])
 263             self.average_first_cycle_intensity = parse_mean_range(data[2])
 264             self.percent_intensity_after_20_cycles = parse_mean_range(data[3])
 265             self.percent_pass_filter_clusters = parse_mean_range(data[4])
 266             self.percent_pass_filter_align = parse_mean_range(data[5])
 267             self.average_alignment_score = parse_mean_range(data[6])
 268             self.percent_error_rate = parse_mean_range(data[7])
 269
 270         def get_elements(self):
 271             lane_result = ElementTree.Element(
 272                             Summary.LaneResultSummary.LANE_RESULT_SUMMARY,
 273                             {'lane': self.lane})
 274             for tag, variable_name in Summary.LaneResultSummary.TAGS.items():
 275                 element = make_mean_range_element(
 276                     lane_result,
 277                     tag,
 278                     *getattr(self, variable_name)
 279                 )
 280             return lane_result
 281
 282         def set_elements(self, tree):
 283             if tree.tag != Summary.LaneResultSummary.LANE_RESULT_SUMMARY:
 284                 raise ValueError('Expected %s' % (
 285                         Summary.LaneResultSummary.LANE_RESULT_SUMMARY))
 286             self.lane = tree.attrib['lane']
 287             tags = Summary.LaneResultSummary.TAGS
 288             for element in list(tree):
 289                 try:
 290                     variable_name = tags[element.tag]
 291                     setattr(self, variable_name,
 292                             parse_mean_range_element(element))
 293                 except KeyError, e:
 294                     logging.warn('Unrecognized tag %s' % (element.tag,))
 295
 296     def __init__(self, filename=None, xml=None):
 297         self.lane_results = {}
 298
 299         if filename is not None:
 300             self._extract_lane_results(filename)
 301         if xml is not None:
 302             self.set_elements(xml)
 303
 304     def __getitem__(self, key):
 305         return self.lane_results[key]
 306
 307     def __len__(self):
 308         return len(self.lane_results)
 309
 310     def keys(self):
 311         return self.lane_results.keys()
 312
 313     def values(self):
 314         return self.lane_results.values()
 315
 316     def items(self):
 317         return self.lane_results.items()
 318
 319     def _extract_lane_results(self, pathname):
 320         """
 321         extract the Lane Results Summary table
 322         """
 323         tree = ElementTree.parse(pathname).getroot()
 324         if flatten(tree.findall('*//h2')[3]) != 'Lane Results Summary':
 325             raise RuntimeError("Summary.htm file format changed")
 326
 327         tables = tree.findall('*//table')
 328
 329         # parse lane result summary
 330         lane_summary = tables[2]
 331         rows = lane_summary.getchildren()
 332         headers = rows[0].getchildren()
 333         if flatten(headers[2]) != 'Av 1st Cycle Int ':
 334             raise RuntimeError("Summary.htm file format changed")
 335
 336         for r in rows[1:]:
 337             lrs = Summary.LaneResultSummary(html=r)
 338             self.lane_results[lrs.lane] = lrs
 339
 340     def get_elements(self):
 341         summary = ElementTree.Element(Summary.SUMMARY,
 342                                       {'version': unicode(Summary.XML_VERSION)})
 343         for lane in self.lane_results.values():
 344             summary.append(lane.get_elements())
 345         return summary
 346
 347     def set_elements(self, tree):
 348         if tree.tag != Summary.SUMMARY:
 349             return ValueError("Expected %s" % (Summary.SUMMARY,))
 350         xml_version = int(tree.attrib.get('version', 0))
 351         if xml_version > Summary.XML_VERSION:
 352             logging.warn('Summary XML tree is a higher version than this class')
 353         for element in list(tree):
 354             lrs = Summary.LaneResultSummary()
 355             lrs.set_elements(element)
 356             self.lane_results[lrs.lane] = lrs
 357
 358     def dump(self):
 359         """
 360         Debugging function, report current object
 361         """
 362         pass
 363
 364
 365 def build_genome_fasta_map(genome_dir):
 366     # build fasta to fasta file map
 367     genome = genome_dir.split(os.path.sep)[-1]
 368     fasta_map = {}
 369     for vld_file in glob(os.path.join(genome_dir, '*.vld')):
 370         is_link = False
 371         if os.path.islink(vld_file):
 372             is_link = True
 373         vld_file = os.path.realpath(vld_file)
 374         path, vld_name = os.path.split(vld_file)
 375         name, ext = os.path.splitext(vld_name)
 376         if is_link:
 377             fasta_map[name] = name
 378         else:
 379             fasta_map[name] = os.path.join(genome, name)
 380     return fasta_map
 381
 382 class ElandLane(object):
 383     """
 384     Process an eland result file
 385     """
 386     XML_VERSION = 1
 387     LANE = 'ElandLane'
 388     SAMPLE_NAME = 'SampleName'
 389     LANE_ID = 'LaneID'
 390     GENOME_MAP = 'GenomeMap'
 391     GENOME_ITEM = 'GenomeItem'
 392     MAPPED_READS = 'MappedReads'
 393     MAPPED_ITEM = 'MappedItem'
 394     MATCH_CODES = 'MatchCodes'
 395     MATCH_ITEM = 'Code'
 396     READS = 'Reads'
 397
 398     def __init__(self, pathname=None, genome_map=None, xml=None):
 399         self.pathname = pathname
 400         self._sample_name = None
 401         self._lane_id = None
 402         self._reads = None
 403         self._mapped_reads = None
 404         self._match_codes = None
 405         if genome_map is None:
 406             genome_map = {}
 407         self.genome_map = genome_map
 408
 409         if xml is not None:
 410             self.set_elements(xml)
 411
 412     def _update(self):
 413         """
 414         Actually read the file and actually count the reads
 415         """
 416         # can't do anything if we don't have a file to process
 417         if self.pathname is None:
 418             return
 419
 420         if os.stat(self.pathname)[stat.ST_SIZE] == 0:
 421             raise RuntimeError("Eland isn't done, try again later.")
 422
 423         reads = 0
 424         mapped_reads = {}
 425
 426         match_codes = {'NM':0, 'QC':0, 'RM':0,
 427                        'U0':0, 'U1':0, 'U2':0,
 428                        'R0':0, 'R1':0, 'R2':0,
 429                       }
 430         for line in autoopen(self.pathname,'r'):
 431             reads += 1
 432             fields = line.split()
 433             # code = fields[2]
 434             # match_codes[code] = match_codes.setdefault(code, 0) + 1
 435             # the QC/NM etc codes are in the 3rd field and always present
 436             match_codes[fields[2]] += 1
 437             # ignore lines that don't have a fasta filename
 438             if len(fields) < 7:
 439                 continue
 440             fasta = self.genome_map.get(fields[6], fields[6])
 441             mapped_reads[fasta] = mapped_reads.setdefault(fasta, 0) + 1
 442         self._match_codes = match_codes
 443         self._mapped_reads = mapped_reads
 444         self._reads = reads
 445
 446     def _update_name(self):
 447         # extract the sample name
 448         if self.pathname is None:
 449             return
 450
 451         path, name = os.path.split(self.pathname)
 452         split_name = name.split('_')
 453         self._sample_name = split_name[0]
 454         self._lane_id = split_name[1]
 455
 456     def _get_sample_name(self):
 457         if self._sample_name is None:
 458             self._update_name()
 459         return self._sample_name
 460     sample_name = property(_get_sample_name)
 461
 462     def _get_lane_id(self):
 463         if self._lane_id is None:
 464             self._update_name()
 465         return self._lane_id
 466     lane_id = property(_get_lane_id)
 467
 468     def _get_reads(self):
 469         if self._reads is None:
 470             self._update()
 471         return self._reads
 472     reads = property(_get_reads)
 473
 474     def _get_mapped_reads(self):
 475         if self._mapped_reads is None:
 476             self._update()
 477         return self._mapped_reads
 478     mapped_reads = property(_get_mapped_reads)
 479
 480     def _get_match_codes(self):
 481         if self._match_codes is None:
 482             self._update()
 483         return self._match_codes
 484     match_codes = property(_get_match_codes)
 485
 486     def get_elements(self):
 487         lane = ElementTree.Element(ElandLane.LANE,
 488                                    {'version':
 489                                     unicode(ElandLane.XML_VERSION)})
 490         sample_tag = ElementTree.SubElement(lane, ElandLane.SAMPLE_NAME)
 491         sample_tag.text = self.sample_name
 492         lane_tag = ElementTree.SubElement(lane, ElandLane.LANE_ID)
 493         lane_tag.text = self.lane_id
 494         genome_map = ElementTree.SubElement(lane, ElandLane.GENOME_MAP)
 495         for k, v in self.genome_map.items():
 496             item = ElementTree.SubElement(
 497                 genome_map, ElandLane.GENOME_ITEM,
 498                 {'name':k, 'value':unicode(v)})
 499         mapped_reads = ElementTree.SubElement(lane, ElandLane.MAPPED_READS)
 500         for k, v in self.mapped_reads.items():
 501             item = ElementTree.SubElement(
 502                 mapped_reads, ElandLane.MAPPED_ITEM,
 503                 {'name':k, 'value':unicode(v)})
 504         match_codes = ElementTree.SubElement(lane, ElandLane.MATCH_CODES)
 505         for k, v in self.match_codes.items():
 506             item = ElementTree.SubElement(
 507                 match_codes, ElandLane.MATCH_ITEM,
 508                 {'name':k, 'value':unicode(v)})
 509         reads = ElementTree.SubElement(lane, ElandLane.READS)
 510         reads.text = unicode(self.reads)
 511
 512         return lane
 513
 514     def set_elements(self, tree):
 515         if tree.tag != ElandLane.LANE:
 516             raise ValueError('Exptecting %s' % (ElandLane.LANE,))
 517
 518         # reset dictionaries
 519         self._mapped_reads = {}
 520         self._match_codes = {}
 521
 522         for element in tree:
 523             tag = element.tag.lower()
 524             if tag == ElandLane.SAMPLE_NAME.lower():
 525                 self._sample_name = element.text
 526             elif tag == ElandLane.LANE_ID.lower():
 527                 self._lane_id = element.text
 528             elif tag == ElandLane.GENOME_MAP.lower():
 529                 for child in element:
 530                     name = child.attrib['name']
 531                     value = child.attrib['value']
 532                     self.genome_map[name] = value
 533             elif tag == ElandLane.MAPPED_READS.lower():
 534                 for child in element:
 535                     name = child.attrib['name']
 536                     value = child.attrib['value']
 537                     self._mapped_reads[name] = int(value)
 538             elif tag == ElandLane.MATCH_CODES.lower():
 539                 for child in element:
 540                     name = child.attrib['name']
 541                     value = int(child.attrib['value'])
 542                     self._match_codes[name] = value
 543             elif tag == ElandLane.READS.lower():
 544                 self._reads = int(element.text)
 545             else:
 546                 logging.warn("ElandLane unrecognized tag %s" % (element.tag,))
 547
 548 def extract_eland_sequence(instream, outstream, start, end):
 549     """
 550     Extract a chunk of sequence out of an eland file
 551     """
 552     for line in instream:
 553         record = line.split()
 554         if len(record) > 1:
 555             result = [record[0], record[1][start:end]]
 556         else:
 557             result = [record[0][start:end]]
 558         outstream.write("\t".join(result))
 559         outstream.write(os.linesep)
 560
 561 class ELAND(object):
 562     """
 563     Summarize information from eland files
 564     """
 565     XML_VERSION = 1
 566
 567     ELAND = 'ElandCollection'
 568     LANE = 'Lane'
 569     LANE_ID = 'id'
 570
 571     def __init__(self, xml=None):
 572         # we need information from the gerald config.xml
 573         self.results = {}
 574
 575         if xml is not None:
 576             self.set_elements(xml)
 577
 578     def __len__(self):
 579         return len(self.results)
 580
 581     def keys(self):
 582         return self.results.keys()
 583
 584     def values(self):
 585         return self.results.values()
 586
 587     def items(self):
 588         return self.results.items()
 589
 590     def __getitem__(self, key):
 591         return self.results[key]
 592
 593     def get_elements(self):
 594         root = ElementTree.Element(ELAND.ELAND,
 595                                    {'version': unicode(ELAND.XML_VERSION)})
 596         for lane_id, lane in self.results.items():
 597             eland_lane = lane.get_elements()
 598             eland_lane.attrib[ELAND.LANE_ID] = unicode(lane_id)
 599             root.append(eland_lane)
 600         return root
 601
 602     def set_elements(self, tree):
 603         if tree.tag.lower() != ELAND.ELAND.lower():
 604             raise ValueError('Expecting %s', ELAND.ELAND)
 605         for element in list(tree):
 606             lane_id = element.attrib[ELAND.LANE_ID]
 607             lane = ElandLane(xml=element)
 608             self.results[lane_id] = lane
 609
 610 def eland(basedir, gerald=None, genome_maps=None):
 611     e = ELAND()
 612
 613     file_list = glob(os.path.join(basedir, "*_eland_result.txt"))
 614     if len(file_list) == 0:
 615         # lets handle compressed eland files too
 616         file_list = glob(os.path.join(basedir, "*_eland_result.txt.bz2"))
 617
 618     for pathname in file_list:
 619         # yes the lane_id is also being computed in ElandLane._update
 620         # I didn't want to clutter up my constructor
 621         # but I needed to persist the sample_name/lane_id for
 622         # runfolder summary_report
 623         path, name = os.path.split(pathname)
 624         split_name = name.split('_')
 625         lane_id = split_name[1]
 626
 627         if genome_maps is not None:
 628             genome_map = genome_maps[lane_id]
 629         elif gerald is not None:
 630             genome_dir = gerald.lanes[lane_id].eland_genome
 631             genome_map = build_genome_fasta_map(genome_dir)
 632         else:
 633             genome_map = {}
 634
 635         eland_result = ElandLane(pathname, genome_map)
 636         e.results[lane_id] = eland_result
 637     return e