htsworkflow/frontend/samples/views.py

   1 # Create your views here.
   2 import StringIO
   3 import logging
   4 import os
   5 import sys
   6
   7 try:
   8     import json
   9 except ImportError, e:
  10     import simplejson as json
  11
  12 from htsworkflow.frontend.auth import require_api_key
  13 from htsworkflow.frontend.experiments.models import FlowCell, Lane
  14 from htsworkflow.frontend.samples.changelist import ChangeList
  15 from htsworkflow.frontend.samples.models import Library, HTSUser
  16 from htsworkflow.frontend.samples.results import get_flowcell_result_dict, parse_flowcell_id
  17 from htsworkflow.frontend.bcmagic.forms import BarcodeMagicForm
  18 from htsworkflow.pipelines.runfolder import load_pipeline_run_xml
  19 from htsworkflow.pipelines import runfolder
  20 from htsworkflow.pipelines.eland import ResultLane
  21 from htsworkflow.frontend import settings
  22 from htsworkflow.util.conversion import unicode_or_none
  23 from htsworkflow.util import makebed
  24 from htsworkflow.util import opener
  25
  26
  27 from django.core.exceptions import ObjectDoesNotExist
  28 from django.http import HttpResponse, HttpResponseRedirect, Http404
  29 from django.shortcuts import render_to_response
  30 from django.template import RequestContext
  31 from django.template.loader import get_template
  32 from django.contrib.auth.decorators import login_required
  33
  34 LANE_LIST = [1,2,3,4,5,6,7,8]
  35 SAMPLES_CONTEXT_DEFAULTS = {
  36     'app_name': 'Flowcell/Library Tracker',
  37     'bcmagic': BarcodeMagicForm()
  38 }
  39
  40 def count_lanes(lane_set):
  41     single = 0
  42     paired = 1
  43     short_read = 0
  44     medium_read = 1
  45     long_read = 2
  46     counts = [[0,0,0,],[0,0,0]]
  47
  48     for lane in lane_set.all():
  49         if lane.flowcell.paired_end:
  50             lane_type = paired
  51         else:
  52             lane_type = single
  53         if lane.flowcell.read_length < 40:
  54             read_type = short_read
  55         elif lane.flowcell.read_length < 100:
  56             read_type = medium_read
  57         else:
  58             read_type = long_read
  59         counts[lane_type][read_type] += 1
  60
  61     return counts
  62
  63 def create_library_context(cl):
  64     """
  65     Create a list of libraries that includes how many lanes were run
  66     """
  67     records = []
  68     #for lib in library_items.object_list:
  69     for lib in cl.result_list:
  70        summary = {}
  71        summary['library_id'] = lib.id
  72        summary['library_name'] = lib.library_name
  73        summary['species_name' ] = lib.library_species.scientific_name
  74        if lib.amplified_from_sample is not None:
  75            summary['amplified_from'] = lib.amplified_from_sample.id
  76        else:
  77            summary['amplified_from'] = ''
  78        lanes_run = count_lanes(lib.lane_set)
  79        # suppress zeros
  80        for row in xrange(len(lanes_run)):
  81            for col in xrange(len(lanes_run[row])):
  82                if lanes_run[row][col] == 0:
  83                    lanes_run[row][col] = ''
  84        summary['lanes_run'] = lanes_run
  85        summary['is_archived'] = lib.is_archived()
  86        records.append(summary)
  87     cl.result_count = unicode(cl.paginator._count) + u" libraries"
  88     return {'library_list': records }
  89
  90 def library(request):
  91    # build changelist
  92     fcl = ChangeList(request, Library,
  93         list_filter=['affiliations', 'library_species'],
  94         search_fields=['id', 'library_name', 'amplified_from_sample__id'],
  95         list_per_page=200,
  96         queryset=Library.objects.filter(hidden__exact=0)
  97     )
  98
  99     context = { 'cl': fcl, 'title': 'Library Index'}
 100     context.update(create_library_context(fcl))
 101     t = get_template('samples/library_index.html')
 102     c = RequestContext(request, context)
 103
 104     app_context = {
 105         'page_name': 'Library Index',
 106         'east_region_config_div': 'changelist-filter',
 107         'body': t.render(c)
 108     }
 109     app_context.update(SAMPLES_CONTEXT_DEFAULTS)
 110
 111     app_t = get_template('flowcell_libraries_app.html')
 112     app_c = RequestContext(request, app_context)
 113     return HttpResponse( app_t.render(app_c) )
 114
 115 def library_to_flowcells(request, lib_id):
 116     """
 117     Display information about all the flowcells a library has been run on.
 118     """
 119
 120     try:
 121       lib = Library.objects.get(id=lib_id)
 122     except:
 123       return HttpResponse("Library %s does not exist" % (lib_id))
 124
 125     flowcell_list = []
 126     flowcell_run_results = {} # aka flowcells we're looking at
 127     for lane in lib.lane_set.all():
 128         fc = lane.flowcell
 129         flowcell_id, id = parse_flowcell_id(fc.flowcell_id)
 130         if flowcell_id not in flowcell_run_results:
 131             flowcell_run_results[flowcell_id] = get_flowcell_result_dict(flowcell_id)
 132         flowcell_list.append((fc.flowcell_id, lane.lane_number))
 133
 134     flowcell_list.sort()
 135     lane_summary_list = []
 136     eland_results = []
 137     for fc, lane_number in flowcell_list:
 138         lane_summary, err_list = _summary_stats(fc, lane_number)
 139
 140         eland_results.extend(_make_eland_results(fc, lane_number, flowcell_run_results))
 141         lane_summary_list.extend(lane_summary)
 142
 143     context = {
 144         'page_name': 'Library Details',
 145         'lib': lib,
 146         'eland_results': eland_results,
 147         'lane_summary_list': lane_summary_list,
 148     }
 149     context.update(SAMPLES_CONTEXT_DEFAULTS)
 150
 151     return render_to_response(
 152         'samples/library_detail.html',
 153         context,
 154         context_instance = RequestContext(request))
 155
 156 def lanes_for(request, username=None):
 157     """
 158     Generate a report of recent activity for a user
 159     """
 160     query = {}
 161     if username is not None:
 162         user = HTSUser.objects.get(username=username)
 163         query.update({'library__affiliations__users__id':user.id})
 164     print query, username
 165     fcl = ChangeList(request, Lane,
 166         list_filter=[],
 167         search_fields=['flowcell__flowcell_id', 'library__id', 'library__library_name'],
 168         list_per_page=200,
 169         queryset=Lane.objects.filter(**query)
 170     )
 171
 172     context = { 'lanes': fcl, 'title': 'Lane Index'}
 173
 174     return render_to_response(
 175         'samples/lanes_for.html',
 176         context,
 177         context_instance = RequestContext(request)
 178     )
 179
 180
 181 def summaryhtm_fc_cnm(request, flowcell_id, cnm):
 182     """
 183     returns a Summary.htm file if it exists.
 184     """
 185     fc_id, status = parse_flowcell_id(flowcell_id)
 186     d = get_flowcell_result_dict(fc_id)
 187
 188     if d is None:
 189         return HttpResponse('<b>Results for Flowcell %s not found.</b>' % (fc_id))
 190
 191     if cnm not in d:
 192         return HttpResponse('<b>Results for Flowcell %s; %s not found.</b>' % (fc_id, cnm))
 193
 194     summary_filepath = d[cnm]['summary']
 195
 196     if summary_filepath is None:
 197         return HttpResponse('<b>Summary.htm for Flowcell %s; %s not found.</b>' % (fc_id, cnm))
 198
 199     f = open(summary_filepath, 'r')
 200
 201     return HttpResponse(f)
 202
 203
 204 def result_fc_cnm_eland_lane(request, flowcell_id, cnm, lane):
 205     """
 206     returns an eland_file upon calling.
 207     """
 208     fc_id, status = parse_flowcell_id(flowcell_id)
 209     d = get_flowcell_result_dict(fc_id)
 210
 211     if d is None:
 212         return HttpResponse('<b>Results for Flowcell %s not found.</b>' % (fc_id))
 213
 214     if cnm not in d:
 215         return HttpResponse('<b>Results for Flowcell %s; %s not found.</b>' % (fc_id, cnm))
 216
 217     erd = d[cnm]['eland_results']
 218     lane = int(lane)
 219
 220     if lane not in erd:
 221         return HttpResponse('<b>Results for Flowcell %s; %s; lane %s not found.</b>' % (fc_id, cnm, lane))
 222
 223     filepath = erd[lane]
 224
 225     #f = opener.autoopen(filepath, 'r')
 226     # return HttpResponse(f, mimetype="application/x-elandresult")
 227
 228     f = open(filepath, 'r')
 229     return HttpResponse(f, mimetype='application/x-bzip2')
 230
 231
 232
 233 def bedfile_fc_cnm_eland_lane_ucsc(request, fc_id, cnm, lane):
 234     """
 235     returns a bed file for a given flowcell, CN-M (i.e. C1-33), and lane (ucsc compatible)
 236     """
 237     return bedfile_fc_cnm_eland_lane(request, fc_id, cnm, lane, ucsc_compatible=True)
 238
 239
 240 def bedfile_fc_cnm_eland_lane(request, flowcell_id, cnm, lane, ucsc_compatible=False):
 241     """
 242     returns a bed file for a given flowcell, CN-M (i.e. C1-33), and lane
 243     """
 244     fc_id, status = parse_flowcell_id(flowcell_id)
 245     d = get_flowcell_result_dict(fc_id)
 246
 247     if d is None:
 248         return HttpResponse('<b>Results for Flowcell %s not found.</b>' % (fc_id))
 249
 250     if cnm not in d:
 251         return HttpResponse('<b>Results for Flowcell %s; %s not found.</b>' % (fc_id, cnm))
 252
 253     erd = d[cnm]['eland_results']
 254     lane = int(lane)
 255
 256     if lane not in erd:
 257         return HttpResponse('<b>Results for Flowcell %s; %s; lane %s not found.</b>' % (fc_id, cnm, lane))
 258
 259     filepath = erd[lane]
 260
 261     # Eland result file
 262     fi = opener.autoopen(filepath, 'r')
 263     # output memory file
 264
 265     name, description = makebed.make_description( fc_id, lane )
 266
 267     bedgen = makebed.make_bed_from_eland_generator(fi, name, description)
 268
 269     if ucsc_compatible:
 270         return HttpResponse(bedgen)
 271     else:
 272         return HttpResponse(bedgen, mimetype="application/x-bedfile")
 273
 274
 275 def _summary_stats(flowcell_id, lane_id):
 276     """
 277     Return the summary statistics for a given flowcell, lane, and end.
 278     """
 279     fc_id, status = parse_flowcell_id(flowcell_id)
 280     fc_result_dict = get_flowcell_result_dict(fc_id)
 281
 282     summary_list = []
 283     err_list = []
 284
 285     if fc_result_dict is None:
 286         err_list.append('Results for Flowcell %s not found.' % (fc_id))
 287         return (summary_list, err_list)
 288
 289     for cycle_width in fc_result_dict:
 290         xmlpath = fc_result_dict[cycle_width]['run_xml']
 291
 292         if xmlpath is None:
 293             err_list.append('Run xml for Flowcell %s(%s) not found.' % (fc_id, cycle_width))
 294             continue
 295
 296         run = load_pipeline_run_xml(xmlpath)
 297         gerald_summary = run.gerald.summary.lane_results
 298         for end in range(len(gerald_summary)):
 299             end_summary = run.gerald.eland_results.results[end]
 300             if end_summary.has_key(lane_id):
 301                 eland_summary = run.gerald.eland_results.results[end][lane_id]
 302             else:
 303                 eland_summary = ResultLane(lane_id=lane_id, end=end)
 304             # add information to lane_summary
 305             eland_summary.flowcell_id = flowcell_id
 306             if len(gerald_summary) > end and gerald_summary[end].has_key(lane_id):
 307                 eland_summary.clusters = gerald_summary[end][lane_id].cluster
 308             else:
 309                 eland_summary.clusters = None
 310             eland_summary.cycle_width = cycle_width
 311             if hasattr(eland_summary, 'genome_map'):
 312                 eland_summary.summarized_reads = runfolder.summarize_mapped_reads(
 313                                                    eland_summary.genome_map,
 314                                                    eland_summary.mapped_reads)
 315
 316             # grab some more information out of the flowcell db
 317             flowcell = FlowCell.objects.get(flowcell_id=flowcell_id)
 318             #pm_field = 'lane_%d_pM' % (lane_id)
 319             lane_obj = flowcell.lane_set.get(lane_number=lane_id)
 320             eland_summary.successful_pm = lane_obj.pM
 321
 322             summary_list.append(eland_summary)
 323
 324         #except Exception, e:
 325         #    summary_list.append("Summary report needs to be updated.")
 326         #    logging.error("Exception: " + str(e))
 327
 328     return (summary_list, err_list)
 329
 330 def _summary_stats_old(flowcell_id, lane):
 331     """
 332     return a dictionary of summary stats for a given flowcell_id & lane.
 333     """
 334     fc_id, status = parse_flowcell_id(flowcell_id)
 335     fc_result_dict = get_flowcell_result_dict(fc_id)
 336
 337     dict_list = []
 338     err_list = []
 339     summary_list = []
 340
 341     if fc_result_dict is None:
 342         err_list.append('Results for Flowcell %s not found.' % (fc_id))
 343         return (dict_list, err_list, summary_list)
 344
 345     for cnm in fc_result_dict:
 346
 347         xmlpath = fc_result_dict[cnm]['run_xml']
 348
 349         if xmlpath is None:
 350             err_list.append('Run xml for Flowcell %s(%s) not found.' % (fc_id, cnm))
 351             continue
 352
 353         tree = ElementTree.parse(xmlpath).getroot()
 354         results = runfolder.PipelineRun(pathname='', xml=tree)
 355         try:
 356             lane_report = runfolder.summarize_lane(results.gerald, lane)
 357             summary_list.append(os.linesep.join(lane_report))
 358         except Exception, e:
 359             summary_list.append("Summary report needs to be updated.")
 360             logging.error("Exception: " + str(e))
 361
 362         print >>sys.stderr, "----------------------------------"
 363         print >>sys.stderr, "-- DOES NOT SUPPORT PAIRED END ---"
 364         print >>sys.stderr, "----------------------------------"
 365         lane_results = results.gerald.summary[0][lane]
 366         lrs = lane_results
 367
 368         d = {}
 369
 370         d['average_alignment_score'] = lrs.average_alignment_score
 371         d['average_first_cycle_intensity'] = lrs.average_first_cycle_intensity
 372         d['cluster'] = lrs.cluster
 373         d['lane'] = lrs.lane
 374         d['flowcell'] = flowcell_id
 375         d['cnm'] = cnm
 376         d['percent_error_rate'] = lrs.percent_error_rate
 377         d['percent_intensity_after_20_cycles'] = lrs.percent_intensity_after_20_cycles
 378         d['percent_pass_filter_align'] = lrs.percent_pass_filter_align
 379         d['percent_pass_filter_clusters'] = lrs.percent_pass_filter_clusters
 380
 381         #FIXME: function finished, but need to take advantage of
 382         #   may need to take in a list of lanes so we only have to
 383         #   load the xml file once per flowcell rather than once
 384         #   per lane.
 385         dict_list.append(d)
 386
 387     return (dict_list, err_list, summary_list)
 388
 389
 390 def get_eland_result_type(pathname):
 391     """
 392     Guess the eland result file type from the filename
 393     """
 394     path, filename = os.path.split(pathname)
 395     if 'extended' in filename:
 396         return 'extended'
 397     elif 'multi' in filename:
 398         return 'multi'
 399     elif 'result' in filename:
 400         return 'result'
 401     else:
 402         return 'unknown'
 403
 404 def _make_eland_results(flowcell_id, lane, interesting_flowcells):
 405     fc_id, status = parse_flowcell_id(flowcell_id)
 406     cur_fc = interesting_flowcells.get(fc_id, None)
 407     if cur_fc is None:
 408       return []
 409
 410     flowcell = FlowCell.objects.get(flowcell_id=flowcell_id)
 411     # Loop throw storage devices if a result has been archived
 412     storage_id_list = []
 413     if cur_fc is not None:
 414         for lts in flowcell.longtermstorage_set.all():
 415             for sd in lts.storage_devices.all():
 416                 # Use barcode_id if it exists
 417                 if sd.barcode_id is not None and sd.barcode_id != '':
 418                     storage_id_list.append(sd.barcode_id)
 419                 # Otherwise use UUID
 420                 else:
 421                     storage_id_list.append(sd.uuid)
 422
 423     # Formatting for template use
 424     if len(storage_id_list) == 0:
 425         storage_ids = None
 426     else:
 427         storage_ids = ', '.join([ '<a href="/inventory/%s/">%s</a>' % (s,s) for s in storage_id_list ])
 428
 429     results = []
 430     for cycle in cur_fc.keys():
 431         result_path = cur_fc[cycle]['eland_results'].get(lane, None)
 432         result_link = make_result_link(fc_id, cycle, lane, result_path)
 433         results.append({'flowcell_id': fc_id,
 434                         'run_date': flowcell.run_date,
 435                         'cycle': cycle,
 436                         'lane': lane,
 437                         'summary_url': make_summary_url(flowcell_id, cycle),
 438                         'result_url': result_link[0],
 439                         'result_label': result_link[1],
 440                         'bed_url': result_link[2],
 441                         'storage_ids': storage_ids
 442         })
 443     return results
 444
 445 def make_summary_url(flowcell_id, cycle_name):
 446     url = '/results/%s/%s/summary/' % (flowcell_id, cycle_name)
 447     return url
 448
 449 def make_result_link(flowcell_id, cycle_name, lane, eland_result_path):
 450     if eland_result_path is None:
 451         return ("", "", "")
 452
 453     result_type = get_eland_result_type(eland_result_path)
 454     result_url = '/results/%s/%s/eland_result/%s' % (flowcell_id, cycle_name, lane)
 455     result_label = 'eland %s' % (result_type,)
 456     bed_url = None
 457     if result_type == 'result':
 458        bed_url_pattern = '/results/%s/%s/bedfile/%s'
 459        bed_url = bed_url_pattern % (flowcell_id, cycle_name, lane)
 460
 461     return (result_url, result_label, bed_url)
 462
 463 def _files(flowcell_id, lane):
 464     """
 465     Sets up available files for download
 466     """
 467     lane = int(lane)
 468
 469     flowcell_id, id = parse_flowcell_id(flowcell_id)
 470     d = get_flowcell_result_dict(flowcell_id)
 471
 472     if d is None:
 473         return ''
 474
 475     output = []
 476
 477     # c_name == 'CN-M' (i.e. C1-33)
 478     for c_name in d:
 479
 480         if d[c_name]['summary'] is not None:
 481             output.append('<a href="/results/%s/%s/summary/">summary(%s)</a>' \
 482                           % (flowcell_id, c_name, c_name))
 483
 484         erd = d[c_name]['eland_results']
 485         if lane in erd:
 486             result_type = get_eland_result_type(erd[lane])
 487             result_url_pattern = '<a href="/results/%s/%s/eland_result/%s">eland %s(%s)</a>'
 488             output.append(result_url_pattern % (flowcell_id, c_name, lane, result_type, c_name))
 489             if result_type == 'result':
 490                 bed_url_pattern = '<a href="/results/%s/%s/bedfile/%s">bedfile(%s)</a>'
 491                 output.append(bed_url_pattern % (flowcell_id, c_name, lane, c_name))
 492
 493     if len(output) == 0:
 494         return ''
 495
 496     return '(' + '|'.join(output) + ')'
 497
 498 def library_id_to_admin_url(request, lib_id):
 499     lib = Library.objects.get(id=lib_id)
 500     return HttpResponseRedirect('/admin/samples/library/%s' % (lib.id,))
 501
 502 def library_dict(library_id):
 503     """
 504     Given a library id construct a dictionary containing important information
 505     return None if nothing was found
 506     """
 507     try:
 508         lib = Library.objects.get(id = library_id)
 509     except Library.DoesNotExist, e:
 510         return None
 511
 512     #lane_info = lane_information(lib.lane_set)
 513     lane_info = []
 514     for lane in lib.lane_set.all():
 515         lane_info.append( {'flowcell':lane.flowcell.flowcell_id,
 516                            'lane_number': lane.lane_number} )
 517
 518     info = {
 519         # 'affiliations'?
 520         # 'aligned_reads': lib.aligned_reads,
 521         #'amplified_into_sample': lib.amplified_into_sample, # into is a colleciton...
 522         #'amplified_from_sample_id': lib.amplified_from_sample,
 523         #'antibody_name': lib.antibody_name(), # we have no antibodies.
 524         'antibody_id': lib.antibody_id,
 525         'avg_lib_size': lib.avg_lib_size,
 526         'cell_line_id': lib.cell_line_id,
 527         'cell_line': unicode_or_none(lib.cell_line),
 528         'experiment_type': lib.experiment_type.name,
 529         'experiment_type_id': lib.experiment_type_id,
 530         'id': lib.id,
 531         'lane_set': lane_info,
 532         'library_id': lib.id,
 533         'library_name': lib.library_name,
 534         'library_species': lib.library_species.scientific_name,
 535         'library_species_id': lib.library_species_id,
 536         #'library_type': lib.library_type.name,
 537         'library_type_id': lib.library_type_id,
 538         'made_for': lib.made_for,
 539         'made_by': lib.made_by,
 540         'notes': lib.notes,
 541         'replicate': lib.replicate,
 542         'stopping_point': lib.stopping_point,
 543         'successful_pM': unicode_or_none(lib.successful_pM),
 544         'undiluted_concentration': unicode_or_none(lib.undiluted_concentration)
 545         }
 546     if lib.library_type_id is None:
 547         info['library_type'] = None
 548     else:
 549         info['library_type'] = lib.library_type.name
 550     return info
 551
 552 def library_json(request, library_id):
 553     """
 554     Return a json formatted library dictionary
 555     """
 556     require_api_key(request)
 557     # what validation should we do on library_id?
 558
 559     lib = library_dict(library_id)
 560     if lib is None:
 561         raise Http404
 562
 563     lib_json = json.dumps(lib)
 564     return HttpResponse(lib_json, mimetype='application/json')
 565
 566 def species_json(request, species_id):
 567     """
 568     Return information about a species.
 569     """
 570     raise Http404
 571
 572 @login_required
 573 def user_profile(request):
 574     """
 575     Information about the user
 576     """
 577     context = {
 578                 'page_name': 'User Profile',
 579                 'media': '',
 580                 #'bcmagic': BarcodeMagicForm(),
 581                 #'select': 'settings',
 582             }
 583     context.update(SAMPLES_CONTEXT_DEFAULTS)
 584     return render_to_response('registration/profile.html', context,
 585                               context_instance=RequestContext(request))
 586