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