htsworkflow/submission/trackhub_submission.py

   1 import logging
   2 import os
   3 from pprint import pformat
   4 import string
   5 import re
   6
   7 import RDF
   8
   9 from htsworkflow.submission.submission import Submission
  10
  11 from htsworkflow.util.rdfhelp import \
  12      fromTypedNode, \
  13      geoSoftNS, \
  14      submissionOntology
  15 from htsworkflow.util.url import parse_ssh_url
  16 from htsworkflow.util.ucsc import bigWigInfo
  17
  18 from django.conf import settings
  19 from django.template import Context, loader
  20 from trackhub import default_hub, CompositeTrack, Track, SuperTrack, ViewTrack
  21 from trackhub.track import TRACKTYPES, SubGroupDefinition
  22 from trackhub.helpers import show_rendered_files
  23 from trackhub.upload import upload_track, upload_hub
  24
  25 LOGGER = logging.getLogger(__name__)
  26
  27 class TrackHubSubmission(Submission):
  28     def __init__(self, name, model, baseurl, baseupload, host):
  29         """Create a trackhub based submission
  30
  31         :Parameters:
  32           - `name`: Name of submission
  33           - `model`: librdf model reference
  34           - `baseurl`: web root where trackhub will be hosted
  35           - `baseupload`: filesystem root where trackhub will be hosted
  36           - `host`: hostname for library pages.
  37         """
  38         super(TrackHubSubmission, self).__init__(name, model, host)
  39         if baseurl is None:
  40             raise ValueError("Need a web root to make a track hub")
  41         self.baseurl = os.path.join(baseurl, self.name)
  42         if baseupload:
  43             sshurl = parse_ssh_url(baseupload)
  44             print sshurl
  45             self.user = sshurl.user
  46             self.host = sshurl.host
  47             self.uploadpath =  sshurl.path
  48         else:
  49             self.uploadpath = None
  50
  51     def make_hub_template(self, result_map):
  52         samples = []
  53         for an_analysis in self.analysis_nodes(result_map):
  54             metadata = self.get_sample_metadata(an_analysis)
  55             if len(metadata) == 0:
  56                 errmsg = 'No metadata found for {0}'
  57                 LOGGER.error(errmsg.format(str(an_analysis),))
  58                 continue
  59             elif len(metadata) > 1:
  60                 errmsg = 'Confused there are more than one sample for %s'
  61                 LOGGER.debug(errmsg % (str(an_analysis),))
  62             metadata = metadata[0]
  63             samples.append(metadata)
  64
  65         template = loader.get_template('trackDb.txt')
  66         context = Context({
  67             'samples': samples,
  68         })
  69         return str(template.render(context))
  70
  71     def make_hub(self, result_map):
  72         genome_db = 'hg19'
  73         hub_url = self.baseurl + '/'
  74         hub, genomes_file, genome, trackdb = default_hub(
  75             hub_name=self.name,
  76             short_label=self.name,
  77             long_label=self.name,
  78             email='email',
  79             genome=genome_db)
  80
  81         hub.remote_dir = self.uploadpath
  82
  83         # build higher order track types
  84         composite = CompositeTrack(
  85             name=self.sanitize_name(self.name),
  86             short_label = self.sanitize_name(self.name),
  87             long_label = str(self.name),
  88             tracktype="bed 3",
  89             dragAndDrop='subtracks',
  90             visibility='full',
  91         )
  92         trackdb.add_tracks(composite)
  93
  94         subgroups = self.add_subgroups(composite)
  95
  96         view_type = None
  97         view = None
  98
  99         for track in self.get_tracks():
 100             if track['file_type'] not in TRACKTYPES:
 101                 LOGGER.info('Unrecognized file type %s', track['file_type'])
 102                 continue
 103
 104             view = self.add_new_view_if_needed(composite, view, track)
 105             track_name = self.make_track_name(track)
 106
 107             track_subgroup = self.make_track_subgroups(subgroups, track)
 108             track_type = self.make_track_type(track)
 109
 110             if 'file_label' in track:
 111                 track_label = self.sanitize_name(track['file_label'])
 112             else:
 113                 track_label = track_name
 114
 115             attributes = {
 116                 'name': track_name,
 117                 'tracktype': track_type,
 118                 'url': hub_url + str(track['relative_path']),
 119                 'short_label': str(track['library_id']),
 120                 'long_label': str(track_label),
 121                 'subgroups': track_subgroup,
 122             }
 123
 124             LOGGER.debug('track attributes: %s', pformat(attributes))
 125             newtrack = Track(**attributes)
 126             view.add_tracks([newtrack])
 127
 128         results = hub.render()
 129         if hub.remote_dir:
 130             LOGGER.info("Uploading to %s @ %s : %s",
 131                         self.user, self.host, hub.remote_dir)
 132             upload_hub(hub=hub, host=self.host, user='diane')
 133
 134     def add_new_view_if_needed(self, composite, view, track):
 135         """Add new trakkhub view if we've hit a new type of track.
 136
 137         :Parameters:
 138           - `composite`: composite track to attach to
 139           - `view_type`: name of view type
 140           - `track`: current track record
 141         """
 142         current_view_type = str(track['output_type'])
 143         if not view or current_view_type != view.name:
 144             attributes = {
 145                 'name': current_view_type,
 146                 'view': current_view_type,
 147                 'visibility': str(track.get('visibility', 'squish')),
 148                 'short_label': current_view_type,
 149                 'tracktype': str(track['file_type'])
 150             }
 151             maxHeightPixels = track.get('maxHeightPixels')
 152             if maxHeightPixels:
 153                 attributes['maxHeightPixels'] = str(maxHeightPixels)
 154             autoScale = track.get('autoScale')
 155             if autoScale:
 156                 attributes['autoScale'] = str(autoScale)
 157             view = ViewTrack(**attributes)
 158             composite.add_view(view)
 159             view_type = current_view_type
 160         return view
 161
 162     def make_manifest(self, result_map):
 163         files = []
 164         for an_analysis in self.analysis_nodes(result_map):
 165             metadata = self.get_manifest_metadata(an_analysis)
 166             files.extend(metadata)
 167
 168         template = loader.get_template('manifest.txt')
 169         context = Context({
 170             'files': files
 171         })
 172         return str(template.render(context))
 173
 174     def make_track_name(self, track):
 175         return '{}_{}_{}'.format(
 176             track['library_id'],
 177             track['replicate'],
 178             track['output_type'],
 179         )
 180
 181     def make_track_subgroups(self, subgroups, track):
 182         track_subgroups = {}
 183         for k in subgroups:
 184             if k in track and track[k]:
 185                 value = self.sanitize_name(track[k])
 186                 track_subgroups[k] = value
 187         return track_subgroups
 188
 189     def make_track_type(self, track):
 190         """Further annotate tracktype.
 191
 192         bigWig files can have additional information. Add it if we can
 193         """
 194         track_type = track['file_type']
 195         if track_type.lower() == 'bigwig':
 196             # something we can enhance
 197             info = bigWigInfo(track['relative_path'])
 198             if info.min is not None and info.max is not None:
 199                 track_type = '{} {} {}'.format(track_type, int(info.min), int(info.max))
 200
 201         LOGGER.debug("track_type: %s", track_type)
 202         return str(track_type)
 203
 204     def add_subgroups(self, composite):
 205         """Add subgroups to composite track"""
 206         search = [ ('htswlib:cell_line', 'cell'),
 207                    ('encode3:rna_type', 'rna_type'),
 208                    ('encode3:protocol', 'protocol'),
 209                    ('htswlib:replicate', 'replicate'),
 210                    ('encode3:library_id', 'library_id'),
 211                    ('encode3:assay', 'assay'),
 212                  ]
 213         subgroups = []
 214         names = []
 215         sortorder = []
 216         dimnames = ('dim{}'.format(x) for x in string.ascii_uppercase)
 217         dimensions = []
 218         filtercomposite = []
 219         for term, name in search:
 220             definitions = self.make_subgroupdefinition(term, name)
 221             if definitions:
 222                 subgroups.append(definitions)
 223                 names.append(name)
 224                 sortorder.append("{}=+".format(name))
 225                 d = dimnames.next()
 226                 dimensions.append("{}={}".format(d, name))
 227                 filtercomposite.append("{}=multi".format(d))
 228
 229         composite.add_subgroups(subgroups)
 230         composite.add_params(sortOrder=' '.join(sortorder))
 231         composite.add_params(dimensions=' '.join(dimensions))
 232         composite.add_params(filterComposite=' '.join(filtercomposite))
 233         return names
 234
 235
 236     def make_subgroupdefinition(self, term, name):
 237         """Subgroup attributes need to be an attribute of the library.
 238         """
 239         template = loader.get_template('trackhub_term_values.sparql')
 240         context = Context({'term': term})
 241         results = self.execute_query(template, context)
 242         values = {}
 243         for row in results:
 244             value = str(row['name'])
 245             values[self.sanitize_name(value)] = value
 246
 247         if values:
 248             return SubGroupDefinition(
 249                     name=name,
 250                     label=name,
 251                     mapping=values,
 252             )
 253         else:
 254             return None
 255
 256     def get_tracks(self):
 257         """Collect information needed to describe trackhub tracks.
 258         """
 259         query_template = loader.get_template('trackhub_samples.sparql')
 260
 261         context = Context({ })
 262
 263         results = self.execute_query(query_template, context)
 264         return results
 265
 266     def sanitize_name(self, name):
 267         replacements = [('poly-?a\+', 'PolyAplus'),
 268                         ('poly-?a-', 'PolyAminus'),
 269                         ('RNA-Seq', 'RNASeq'),
 270                         ('rna-seq', 'rnaseq'),
 271                         ('-', '_'),
 272                         (' ', '_'),
 273                         ('^0', 'Zero'),
 274                         ('^1', 'One'),
 275                         ('^2', 'Two'),
 276                         ('^3', 'Three'),
 277                         ('^4', 'Four'),
 278                         ('^5', 'Five'),
 279                         ('^6', 'Six'),
 280                         ('^7', 'Seven'),
 281                         ('^8', 'Eight'),
 282                         ('^9', 'Nine'),
 283                         ]
 284
 285         for regex, substitution in replacements:
 286             name = re.sub(regex, substitution, name, flags=re.IGNORECASE)
 287
 288         return name
 289
 290     def get_manifest_metadata(self, analysis_node):
 291         query_template = loader.get_template('trackhub_manifest.sparql')
 292
 293         context = Context({
 294             'submission': str(analysis_node.uri),
 295             'submissionSet': str(self.submissionSetNS[''].uri),
 296             })
 297         results = self.execute_query(query_template, context)
 298         LOGGER.info("scanned %s for results found %s",
 299                     str(analysis_node), len(results))
 300         return results