htsworkflow/pipelines/summary.py

   1 """
   2 Analyze the Summary.htm file produced by GERALD
   3 """
   4 from __future__ import print_function, unicode_literals
   5
   6 import os
   7 import logging
   8 import re
   9 from pprint import pprint
  10
  11 from lxml import html
  12 from lxml import etree
  13 from htsworkflow.util.ethelp import indent, flatten
  14
  15 LOGGER = logging.getLogger(__name__)
  16 nan = float('nan')
  17
  18 class Summary(object):
  19     """
  20     Extract some useful information from the Summary.htm file
  21     """
  22     XML_VERSION = 3
  23     SUMMARY = 'Summary'
  24
  25     def __init__(self, filename=None, xml=None):
  26         # lane results is a list of 1 or 2 ends containing
  27         # a dictionary of all the lanes reported in this
  28         # summary file
  29         self.lane_results = [{}]
  30
  31         if filename is not None:
  32             self._extract_lane_results(filename)
  33         if xml is not None:
  34             self.set_elements(xml)
  35
  36     def __getitem__(self, key):
  37         return self.lane_results[key]
  38
  39     def __len__(self):
  40         return len(self.lane_results)
  41
  42     def get_elements(self):
  43         summary = etree.Element(Summary.SUMMARY,
  44                                       {'version': str(Summary.XML_VERSION)})
  45         for end in self.lane_results:
  46             for lane in end.values():
  47                 summary.append(lane.get_elements())
  48         return summary
  49
  50     def set_elements(self, tree):
  51         if tree.tag != Summary.SUMMARY:
  52             return ValueError("Expected %s" % (Summary.SUMMARY,))
  53         xml_version = int(tree.attrib.get('version', 0))
  54         if xml_version > Summary.XML_VERSION:
  55             LOGGER.warn('Summary XML tree is a higher version than this class')
  56         for element in list(tree):
  57             lrs = LaneResultSummaryGA()
  58             lrs.set_elements(element)
  59             if len(self.lane_results) < (lrs.end + 1):
  60               self.lane_results.append({})
  61             self.lane_results[lrs.end][lrs.lane] = lrs
  62
  63     def is_paired_end(self):
  64       return len(self.lane_results) == 2
  65
  66     def dump(self):
  67         """
  68         Debugging function, report current object
  69         """
  70         tree = self.get_elements()
  71         print(etree.tostring(tree))
  72
  73 class SummaryGA(Summary):
  74     def __init__(self, filename=None, xml=None):
  75         super(SummaryGA, self).__init__(filename, xml)
  76
  77     def _flattened_row(self, row):
  78         """
  79         flatten the children of a <tr>...</tr>
  80         """
  81         return [flatten(x) for x in row.getchildren() ]
  82
  83     def _parse_table(self, table):
  84         """
  85         assumes the first line is the header of a table,
  86         and that the remaining rows are data
  87         """
  88         rows = table.getchildren()
  89         data = []
  90         for r in rows:
  91             data.append(self._flattened_row(r))
  92         return data
  93
  94     def _extract_lane_results(self, pathname):
  95         """
  96         Extract just the lane results.
  97         Currently those are the only ones we care about.
  98         """
  99
 100         tables = self._extract_named_tables(pathname)
 101
 102
 103     def _extract_named_tables(self, pathname):
 104         """
 105         extract all the 'named' tables from a Summary.htm file
 106         and return as a dictionary
 107
 108         Named tables are <h2>...</h2><table>...</table> pairs
 109         The contents of the h2 tag is considered to the name
 110         of the table.
 111         """
 112         if isxml_file(pathname):
 113             tree = etree.parse(pathname).getroot()
 114         else:
 115             # html
 116             tree = html.parse(pathname).getroot()
 117         # tree = etree.parse(pathname).getroot()
 118         # hack for 1.1rc1, this should be removed when possible.
 119         #file_body = open(pathname).read()
 120         #file_body = file_body.replace('CHASTITY<=', 'CHASTITY&lt;=')
 121         #tree = etree.fromstring(file_body)
 122
 123         # are we reading the xml or the html version of the Summary file?
 124         if tree.tag.lower() == 'summary':
 125             # summary version
 126             tables = self._extract_named_tables_from_gerald_xml(tree)
 127         elif tree.tag.lower() == 'html':
 128             # html version
 129             tables = self._extract_named_tables_from_html(tree)
 130             table_names = [ ('Lane Results Summary', 0),
 131                             ('Lane Results Summary : Read 1', 0),
 132                             ('Lane Results Summary : Read 2', 1),]
 133             for name, end in table_names:
 134                 if name in tables:
 135                     self._extract_lane_results_for_end(tables, name, end)
 136
 137         if len(self.lane_results[0])  == 0:
 138             LOGGER.warning("No Lane Results Summary Found in %s" % (pathname,))
 139
 140     def _extract_named_tables_from_gerald_xml(self, tree):
 141         """
 142         Extract useful named tables from a gerald created Summary.xml file
 143         """
 144         # using the function to convert to lower instead of just writing it
 145         # makes the tag easier to read (IMO)
 146         useful_tables = ['LaneResultsSummary'.lower(),]
 147
 148         tables ={}
 149         for child in tree.getchildren():
 150             if child.tag.lower() in  useful_tables:
 151                 read_tree = child.find('Read')
 152                 # we want 0 based.
 153                 read = int(read_tree.find('readNumber').text)-1
 154                 for element in read_tree.getchildren():
 155                     if element.tag.lower() == "lane":
 156                         lrs = LaneResultSummaryGA()
 157                         lrs.set_elements_from_gerald_xml(read, element)
 158                         self.lane_results[lrs.end][lrs.lane] = lrs
 159         # probably not useful
 160         return tables
 161
 162     ###### START HTML Table Extraction ########
 163     def _extract_named_tables_from_html(self, tree):
 164         tables = {}
 165         for t in tree.findall('*//table'):
 166             previous = t.getprevious()
 167             if previous is None:
 168                 previous = t.getparent()
 169
 170             if previous.tag.lower() == 'div':
 171                 previous = previous.getprevious()
 172
 173             if previous.tag in ('h2', 'p'):
 174                 # we have a table
 175                 name = flatten(previous)
 176                 data = self._parse_table(t)
 177                 tables[name] = data
 178         return tables
 179
 180     def _extract_lane_results_for_end(self, tables, table_name, end):
 181         """
 182         extract the Lane Results Summary table
 183         """
 184         # parse lane result summary
 185         lane_summary = tables[table_name]
 186         # this is version 1 of the summary file
 187         if len(lane_summary[-1]) == 8:
 188             # strip header
 189             headers = lane_summary[0]
 190             # grab the lane by lane data
 191             lane_summary = lane_summary[1:]
 192
 193         # len(lane_summary[-1] = 10 is version 2 of the summary file
 194         #                      = 9  is version 3 of the Summary.htm file
 195         elif len(lane_summary[-1]) in (9, 10):
 196             # lane_summary[0] is a different less specific header row
 197             headers = lane_summary[1]
 198             lane_summary = lane_summary[2:10]
 199             # after the last lane, there's a set of chip wide averages
 200
 201         # append an extra dictionary if needed
 202         if len(self.lane_results) < (end + 1):
 203           self.lane_results.append({})
 204
 205         for r in lane_summary:
 206             lrs = LaneResultSummaryGA(html=r)
 207             lrs.end = end
 208             self.lane_results[lrs.end][lrs.lane] = lrs
 209     ###### END HTML Table Extraction ########
 210
 211
 212 class SummaryHiSeq(Summary):
 213     def __init__(self, filename=None, xml=None):
 214         super(SummaryHiSeq, self).__init__(filename, xml)
 215
 216     def _extract_lane_results(self, filename):
 217         read1 = os.path.join(filename, 'read1.xml')
 218         read2 = os.path.join(filename, 'read2.xml')
 219
 220         if os.path.exists(read1):
 221             self._extract_lane_results_for_end(read1, 0)
 222         else:
 223             LOGGER.warn("No read1.xml at %s." % (read1,))
 224         if os.path.exists(read2):
 225             self.lane_results.append({})
 226             self._extract_lane_results_for_end(read2, 1)
 227
 228     def _extract_lane_results_for_end(self, filename, end):
 229         self.tree = etree.parse(filename)
 230         root = self.tree.getroot()
 231         for lane in root.getchildren():
 232             lrs = LaneResultSummaryHiSeq(data=lane)
 233             lrs.end = end
 234             self.lane_results[lrs.end][lrs.lane] = lrs
 235
 236
 237 class LaneResultSummary(object):
 238     """
 239     Parse the LaneResultSummary table out of Summary.htm
 240     Mostly for the cluster number
 241     """
 242     LANE_RESULT_SUMMARY = 'LaneResultSummary'
 243     TAGS = {
 244       'LaneYield': 'lane_yield',
 245       'Cluster': 'cluster', # Raw
 246       'ClusterPF': 'cluster_pass_filter',
 247       'AverageFirstCycleIntensity': 'average_first_cycle_intensity',
 248       'PercentIntensityAfter20Cycles': 'percent_intensity_after_20_cycles',
 249       'PercentPassFilterClusters': 'percent_pass_filter_clusters',
 250       'PercentPassFilterAlign': 'percent_pass_filter_align',
 251       'AverageAlignmentScore': 'average_alignment_score',
 252       'PercentErrorRate': 'percent_error_rate'
 253     }
 254     # These are tags that have mean/stdev as found in the GERALD Summary.xml file
 255     GERALD_TAGS = {
 256       #'laneYield': 'lane_yield', #this is just a number
 257       'clusterCountRaw': 'cluster', # Raw
 258       'clusterCountPF': 'cluster_pass_filter',
 259       'oneSig': 'average_first_cycle_intensity',
 260       'signal20AsPctOf1': 'percent_intensity_after_20_cycles',
 261       'percentClustersPF': 'percent_pass_filter_clusters',
 262       'percentUniquelyAlignedPF': 'percent_pass_filter_align',
 263       'averageAlignScorePF': 'average_alignment_score',
 264       'errorPF': 'percent_error_rate'
 265     }
 266
 267     def __init__(self, html=None, xml=None):
 268         self.lane = None
 269         self.end = 0
 270         self.lane_yield = None
 271         self.cluster = None
 272         self.cluster_pass_filter = None
 273         self.average_first_cycle_intensity = None
 274         self.percent_intensity_after_20_cycles = None
 275         self.percent_pass_filter_clusters = None
 276         self.percent_pass_filter_align = None
 277         self.average_alignment_score = None
 278         self.percent_error_rate = None
 279
 280
 281     def get_elements(self):
 282         lane_result = etree.Element(
 283                         LaneResultSummary.LANE_RESULT_SUMMARY,
 284                         {'lane': str(self.lane), 'end': str(self.end)})
 285         for tag, variable_name in LaneResultSummary.TAGS.items():
 286             value = getattr(self, variable_name)
 287             if value is None:
 288                 continue
 289             elif isinstance(value, (tuple, list)):
 290                 element = make_mean_range_element(
 291                   lane_result,
 292                   tag,
 293                   *value
 294                 )
 295             else:
 296                 element = etree.SubElement(lane_result, tag)
 297                 element.text = str(value)
 298         return lane_result
 299
 300     def set_elements(self, tree):
 301         if tree.tag != LaneResultSummary.LANE_RESULT_SUMMARY:
 302             raise ValueError('Expected %s' % (
 303                     LaneResultSummary.LANE_RESULT_SUMMARY))
 304         self.lane = int(tree.attrib['lane'])
 305         # default to the first end, for the older summary files
 306         # that are single ended
 307         self.end = int(tree.attrib.get('end', 0))
 308         tags = LaneResultSummary.TAGS
 309         for element in list(tree):
 310             try:
 311                 variable_name = tags[element.tag]
 312                 setattr(self, variable_name,
 313                         parse_summary_element(element))
 314             except KeyError as e:
 315                 LOGGER.warn('Unrecognized tag %s' % (element.tag,))
 316
 317
 318 class LaneResultSummaryGA(LaneResultSummary):
 319     def __init__(self, html=None, xml=None):
 320         super(LaneResultSummaryGA, self).__init__(html, xml)
 321
 322         if html is not None:
 323             self.set_elements_from_source(html)
 324         if xml is not None:
 325             self.set_elements(xml)
 326
 327     def set_elements_from_gerald_xml(self, read, element):
 328         self.lane = int(element.find('laneNumber').text)
 329         self.end = read
 330         lane_yield_node = element.find('laneYield')
 331         if lane_yield_node is not None:
 332             self.lane_yield = int(lane_yield_node.text)
 333         else:
 334             self.lane_yield = None
 335
 336         for GeraldName, LRSName in LaneResultSummary.GERALD_TAGS.items():
 337             node = element.find(GeraldName)
 338             if node is None:
 339                 LOGGER.info("Couldn't find %s" % (GeraldName))
 340             setattr(self, LRSName, parse_xml_mean_range(node))
 341
 342     def set_elements_from_source(self, data):
 343         """Read from an initial summary data file. Either xml or html
 344         """
 345         if not len(data) in (8,10):
 346             raise RuntimeError("Summary.htm file format changed, len(data)=%d" % (len(data),))
 347
 348         # same in pre-0.3.0 Summary file and 0.3 summary file
 349         self.lane = int(data[0])
 350
 351         if len(data) == 8:
 352             parsed_data = [ parse_mean_range(x) for x in data[1:] ]
 353             # this is the < 0.3 Pipeline version
 354             self.cluster = parsed_data[0]
 355             self.average_first_cycle_intensity = parsed_data[1]
 356             self.percent_intensity_after_20_cycles = parsed_data[2]
 357             self.percent_pass_filter_clusters = parsed_data[3]
 358             self.percent_pass_filter_align = parsed_data[4]
 359             self.average_alignment_score = parsed_data[5]
 360             self.percent_error_rate = parsed_data[6]
 361         elif len(data) == 10:
 362             parsed_data = [ parse_mean_range(x) for x in data[2:] ]
 363             # this is the >= 0.3 summary file
 364             self.lane_yield = data[1]
 365             self.cluster = parsed_data[0]
 366             self.cluster_pass_filter = parsed_data[1]
 367             self.average_first_cycle_intensity = parsed_data[2]
 368             self.percent_intensity_after_20_cycles = parsed_data[3]
 369             self.percent_pass_filter_clusters = parsed_data[4]
 370             self.percent_pass_filter_align = parsed_data[5]
 371             self.average_alignment_score = parsed_data[6]
 372             self.percent_error_rate = parsed_data[7]
 373
 374
 375 class LaneResultSummaryHiSeq(LaneResultSummary):
 376     def __init__(self, data=None, xml=None):
 377         super(LaneResultSummaryHiSeq, self).__init__(data, xml)
 378
 379         if data is not None:
 380             self.set_elements_from_source(data)
 381         if xml is not None:
 382             self.set_elements(xml)
 383
 384     def set_elements_from_source(self, element):
 385         """Read from an initial summary data file. Either xml or html
 386         """
 387         # same in pre-0.3.0 Summary file and 0.3 summary file
 388         lane = element.attrib
 389         self.lane = int(lane['key'])
 390         #self.lane_yield = data[1]
 391         self.cluster = (int(lane['ClustersRaw']),
 392                         float(lane['ClustersRawSD']))
 393         self.cluster_pass_filter = (int(lane['ClustersPF']),
 394                                     float(lane['ClustersPFSD']))
 395         self.average_first_cycle_intensity = (int(lane['FirstCycleIntPF']),
 396                                               float(lane['FirstCycleIntPFSD']))
 397         self.percent_intensity_after_20_cycles = (
 398             float(lane['PrcIntensityAfter20CyclesPF']),
 399             float(lane['PrcIntensityAfter20CyclesPFSD']))
 400         self.percent_pass_filter_clusters = (
 401             float(lane['PrcPFClusters']),
 402             float(lane['PrcPFClustersSD']))
 403         self.percent_pass_filter_align = (
 404             float(lane['PrcAlign']),
 405             float(lane['PrcAlignSD']))
 406         self.percent_error_rate = (
 407             float(lane['ErrRatePhiX']),
 408             float(lane['ErrRatePhiXSD']),)
 409
 410
 411 def tonumber(v):
 412     """
 413     Convert a value to int if its an int otherwise a float.
 414     """
 415     try:
 416         v = int(v)
 417     except ValueError as e:
 418         v = float(v)
 419     return v
 420
 421 def parse_mean_range(value):
 422     """
 423     Parse values like 123 +/- 4.5
 424     """
 425     if value.strip() == 'unknown':
 426         return nan, nan
 427
 428     values = value.split()
 429     if len(values) == 1:
 430         if values[0] == '+/-':
 431             return nan,nan
 432         else:
 433             return tonumber(values[0])
 434
 435     average, pm, deviation = values
 436     if pm != '+/-':
 437         raise RuntimeError("Summary.htm file format changed")
 438     return tonumber(average), tonumber(deviation)
 439
 440 def make_mean_range_element(parent, name, mean, deviation):
 441     """
 442     Make an etree subelement <Name mean='mean', deviation='deviation'/>
 443     """
 444     element = etree.SubElement(parent, name,
 445                                      { 'mean': str(mean),
 446                                        'deviation': str(deviation)})
 447     return element
 448
 449 def parse_mean_range_element(element):
 450     """
 451     Grab mean/deviation out of element
 452     """
 453     return (tonumber(element.attrib['mean']),
 454             tonumber(element.attrib['deviation']))
 455
 456 def parse_summary_element(element):
 457     """
 458     Determine if we have a simple element or a mean/deviation element
 459     """
 460     if len(element.attrib) > 0:
 461         return parse_mean_range_element(element)
 462     else:
 463         return element.text
 464
 465 def parse_xml_mean_range(element):
 466     """
 467     Extract mean/stddev children from an element as a tuple
 468     """
 469     if element is None:
 470         return None
 471
 472     mean = element.find('mean')
 473     stddev = element.find('stdev')
 474     if mean is None or stddev is None:
 475         raise RuntimeError("Summary.xml file format changed, expected mean/stddev tags")
 476     if mean.text is None:
 477         mean_value = float('nan')
 478     else:
 479         mean_value = tonumber(mean.text)
 480
 481     if stddev.text is None:
 482         stddev_value = float('nan')
 483     else:
 484         stddev_value = tonumber(stddev.text)
 485
 486
 487     return (mean_value, stddev_value)
 488
 489 def isxml_file(fname):
 490     with open(fname,'r') as stream:
 491         return isxml_stream(stream)
 492
 493 def isxml_stream(stream):
 494     """Return true or false if its sort of an xml file
 495     """
 496     pos = stream.tell()
 497     line = stream.readline()
 498     stream.seek(pos)
 499     if re.match("<\?xml.*?>", line):
 500         # attempt at xml
 501         return True
 502     else:
 503         return False
 504
 505 if __name__ == "__main__":
 506     # test code
 507     from optparse import OptionParser
 508     parser = OptionParser('%prog [Summary.xml/Summary.htm]+')
 509     opts, args = parser.parse_args()
 510     if len(args) == 0:
 511         parser.error('need at least one xml/html file')
 512     for fname in args:
 513         s = Summary(fname)
 514         s.dump()
 515