htsworkflow/pipelines/summary.py

   1 """
   2 Analyze the Summary.htm file produced by GERALD
   3 """
   4 import types
   5
   6 from htsworkflow.pipelines.runfolder import ElementTree
   7 from htsworkflow.util.ethelp import indent, flatten
   8
   9 class Summary(object):
  10     """
  11     Extract some useful information from the Summary.htm file
  12     """
  13     XML_VERSION = 3
  14     SUMMARY = 'Summary'
  15
  16     class LaneResultSummary(object):
  17         """
  18         Parse the LaneResultSummary table out of Summary.htm
  19         Mostly for the cluster number
  20         """
  21         LANE_RESULT_SUMMARY = 'LaneResultSummary'
  22         TAGS = {
  23           'LaneYield': 'lane_yield',
  24           'Cluster': 'cluster', # Raw
  25           'ClusterPF': 'cluster_pass_filter',
  26           'AverageFirstCycleIntensity': 'average_first_cycle_intensity',
  27           'PercentIntensityAfter20Cycles': 'percent_intensity_after_20_cycles',
  28           'PercentPassFilterClusters': 'percent_pass_filter_clusters',
  29           'PercentPassFilterAlign': 'percent_pass_filter_align',
  30           'AverageAlignmentScore': 'average_alignment_score',
  31           'PercentErrorRate': 'percent_error_rate'
  32         }
  33
  34         def __init__(self, html=None, xml=None):
  35             self.lane = None
  36             self.end = 0
  37             self.lane_yield = None
  38             self.cluster = None
  39             self.cluster_pass_filter = None
  40             self.average_first_cycle_intensity = None
  41             self.percent_intensity_after_20_cycles = None
  42             self.percent_pass_filter_clusters = None
  43             self.percent_pass_filter_align = None
  44             self.average_alignment_score = None
  45             self.percent_error_rate = None
  46
  47             if html is not None:
  48                 self.set_elements_from_html(html)
  49             if xml is not None:
  50                 self.set_elements(xml)
  51
  52         def set_elements_from_html(self, data):
  53             if not len(data) in (8,10):
  54                 raise RuntimeError("Summary.htm file format changed")
  55
  56             # same in pre-0.3.0 Summary file and 0.3 summary file
  57             self.lane = int(data[0])
  58
  59             if len(data) == 8:
  60                 parsed_data = [ parse_mean_range(x) for x in data[1:] ]
  61                 # this is the < 0.3 Pipeline version
  62                 self.cluster = parsed_data[0]
  63                 self.average_first_cycle_intensity = parsed_data[1]
  64                 self.percent_intensity_after_20_cycles = parsed_data[2]
  65                 self.percent_pass_filter_clusters = parsed_data[3]
  66                 self.percent_pass_filter_align = parsed_data[4]
  67                 self.average_alignment_score = parsed_data[5]
  68                 self.percent_error_rate = parsed_data[6]
  69             elif len(data) == 10:
  70                 parsed_data = [ parse_mean_range(x) for x in data[2:] ]
  71                 # this is the >= 0.3 summary file
  72                 self.lane_yield = data[1]
  73                 self.cluster = parsed_data[0]
  74                 self.cluster_pass_filter = parsed_data[1]
  75                 self.average_first_cycle_intensity = parsed_data[2]
  76                 self.percent_intensity_after_20_cycles = parsed_data[3]
  77                 self.percent_pass_filter_clusters = parsed_data[4]
  78                 self.percent_pass_filter_align = parsed_data[5]
  79                 self.average_alignment_score = parsed_data[6]
  80                 self.percent_error_rate = parsed_data[7]
  81
  82         def get_elements(self):
  83             lane_result = ElementTree.Element(
  84                             Summary.LaneResultSummary.LANE_RESULT_SUMMARY,
  85                             {'lane': str(self.lane), 'end': str(self.end)})
  86             for tag, variable_name in Summary.LaneResultSummary.TAGS.items():
  87                 value = getattr(self, variable_name)
  88                 if value is None:
  89                     continue
  90                 # it looks like a sequence
  91                 elif type(value) in (types.TupleType, types.ListType):
  92                     element = make_mean_range_element(
  93                       lane_result,
  94                       tag,
  95                       *value
  96                     )
  97                 else:
  98                     element = ElementTree.SubElement(lane_result, tag)
  99                     element.text = value
 100             return lane_result
 101
 102         def set_elements(self, tree):
 103             if tree.tag != Summary.LaneResultSummary.LANE_RESULT_SUMMARY:
 104                 raise ValueError('Expected %s' % (
 105                         Summary.LaneResultSummary.LANE_RESULT_SUMMARY))
 106             self.lane = int(tree.attrib['lane'])
 107             # default to the first end, for the older summary files
 108             # that are single ended
 109             self.end = int(tree.attrib.get('end', 0))
 110             tags = Summary.LaneResultSummary.TAGS
 111             for element in list(tree):
 112                 try:
 113                     variable_name = tags[element.tag]
 114                     setattr(self, variable_name,
 115                             parse_summary_element(element))
 116                 except KeyError, e:
 117                     logging.warn('Unrecognized tag %s' % (element.tag,))
 118
 119     def __init__(self, filename=None, xml=None):
 120         # lane results is a list of 1 or 2 ends containing
 121         # a dictionary of all the lanes reported in this
 122         # summary file
 123         self.lane_results = [{}]
 124
 125         if filename is not None:
 126             self._extract_lane_results(filename)
 127         if xml is not None:
 128             self.set_elements(xml)
 129
 130     def __getitem__(self, key):
 131         return self.lane_results[key]
 132
 133     def __len__(self):
 134         return len(self.lane_results)
 135
 136     def _flattened_row(self, row):
 137         """
 138         flatten the children of a <tr>...</tr>
 139         """
 140         return [flatten(x) for x in row.getchildren() ]
 141
 142     def _parse_table(self, table):
 143         """
 144         assumes the first line is the header of a table,
 145         and that the remaining rows are data
 146         """
 147         rows = table.getchildren()
 148         data = []
 149         for r in rows:
 150             data.append(self._flattened_row(r))
 151         return data
 152
 153     def _extract_named_tables(self, pathname):
 154         """
 155         extract all the 'named' tables from a Summary.htm file
 156         and return as a dictionary
 157
 158         Named tables are <h2>...</h2><table>...</table> pairs
 159         The contents of the h2 tag is considered to the name
 160         of the table.
 161         """
 162         # tree = ElementTree.parse(pathname).getroot()
 163         # hack for 1.1rc1, this should be removed when possible.
 164         file_body = open(pathname).read()
 165         file_body = file_body.replace('CHASTITY<=', 'CHASTITY&lt;=')
 166         tree = ElementTree.fromstring(file_body)
 167         body = tree.find('body')
 168         tables = {}
 169         for i in range(len(body)):
 170             if body[i].tag == 'h2' and body[i+1].tag == 'table':
 171                 # we have an interesting table
 172                 name = flatten(body[i])
 173                 table = body[i+1]
 174                 data = self._parse_table(table)
 175                 tables[name] = data
 176         return tables
 177
 178     def _extract_lane_results(self, pathname):
 179         tables = self._extract_named_tables(pathname)
 180         table_names = [ ('Lane Results Summary', 0),
 181                         ('Lane Results Summary : Read 1', 0),
 182                         ('Lane Results Summary : Read 2', 1),]
 183         for name, end in table_names:
 184           if tables.has_key(name):
 185             self._extract_lane_results_for_end(tables, name, end)
 186
 187     def _extract_lane_results_for_end(self, tables, table_name, end):
 188         """
 189         extract the Lane Results Summary table
 190         """
 191         # parse lane result summary
 192         lane_summary = tables[table_name]
 193         # this is version 1 of the summary file
 194         if len(lane_summary[-1]) == 8:
 195             # strip header
 196             headers = lane_summary[0]
 197             # grab the lane by lane data
 198             lane_summary = lane_summary[1:]
 199
 200         # this is version 2 of the summary file
 201         if len(lane_summary[-1]) == 10:
 202             # lane_summary[0] is a different less specific header row
 203             headers = lane_summary[1]
 204             lane_summary = lane_summary[2:10]
 205             # after the last lane, there's a set of chip wide averages
 206
 207         # append an extra dictionary if needed
 208         if len(self.lane_results) < (end + 1):
 209           self.lane_results.append({})
 210
 211         for r in lane_summary:
 212             lrs = Summary.LaneResultSummary(html=r)
 213             lrs.end = end
 214             self.lane_results[lrs.end][lrs.lane] = lrs
 215
 216     def get_elements(self):
 217         summary = ElementTree.Element(Summary.SUMMARY,
 218                                       {'version': unicode(Summary.XML_VERSION)})
 219         for end in self.lane_results:
 220             for lane in end.values():
 221                 summary.append(lane.get_elements())
 222         return summary
 223
 224     def set_elements(self, tree):
 225         if tree.tag != Summary.SUMMARY:
 226             return ValueError("Expected %s" % (Summary.SUMMARY,))
 227         xml_version = int(tree.attrib.get('version', 0))
 228         if xml_version > Summary.XML_VERSION:
 229             logging.warn('Summary XML tree is a higher version than this class')
 230         for element in list(tree):
 231             lrs = Summary.LaneResultSummary()
 232             lrs.set_elements(element)
 233             print lrs.end, lrs.lane
 234             if len(self.lane_results) < (lrs.end + 1):
 235               self.lane_results.append({})
 236             self.lane_results[lrs.end][lrs.lane] = lrs
 237
 238     def is_paired_end(self):
 239       return len(self.lane_results) == 2
 240
 241     def dump(self):
 242         """
 243         Debugging function, report current object
 244         """
 245         pass
 246
 247 def tonumber(v):
 248     """
 249     Convert a value to int if its an int otherwise a float.
 250     """
 251     try:
 252         v = int(v)
 253     except ValueError, e:
 254         v = float(v)
 255     return v
 256
 257 def parse_mean_range(value):
 258     """
 259     Parse values like 123 +/- 4.5
 260     """
 261     if value.strip() == 'unknown':
 262         return 0, 0
 263
 264     average, pm, deviation = value.split()
 265     if pm != '+/-':
 266         raise RuntimeError("Summary.htm file format changed")
 267     return tonumber(average), tonumber(deviation)
 268
 269 def make_mean_range_element(parent, name, mean, deviation):
 270     """
 271     Make an ElementTree subelement <Name mean='mean', deviation='deviation'/>
 272     """
 273     element = ElementTree.SubElement(parent, name,
 274                                      { 'mean': unicode(mean),
 275                                        'deviation': unicode(deviation)})
 276     return element
 277
 278 def parse_mean_range_element(element):
 279     """
 280     Grab mean/deviation out of element
 281     """
 282     return (tonumber(element.attrib['mean']),
 283             tonumber(element.attrib['deviation']))
 284
 285 def parse_summary_element(element):
 286     """
 287     Determine if we have a simple element or a mean/deviation element
 288     """
 289     if len(element.attrib) > 0:
 290         return parse_mean_range_element(element)
 291     else:
 292         return element.text