def upload(self, object_ids_f=None, feature_matrix_f=None):
smi = 2
self.fetch_session_data()
if self.project_id is None:
return self.no_project_selected()
pm = self.project_manager
fe = pm.get_feature_extraction()
kw_args = self.get_template_args(smi)
kw_args["fe"] = fe
# upload custom feature matrix
error_msg = None
if object_ids_f and feature_matrix_f:
if object_ids_f.file is None:
error_msg = "No protein ids file selected"
elif feature_matrix_f.file is None:
error_msg = "No feature matrix file selected"
else:
error_msg = pm.add_custom_features(self.project_id, object_ids_f, feature_matrix_f)
if error_msg == "":
# redirect to feature list if no errors occured
raise cherrypy.HTTPRedirect(self.get_url(0))
else:
kw_args["msg"] = error_msg
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)
def details(self, cl_id):
smi = 2
self.fetch_session_data()
pm = self.project_manager
if(self.project_id is None):
return self.no_project_selected()
if not(cl_id in pm.get_classifier_ids()):
return self.no_such_classifier()
kw_args = self.get_template_args(smi)
kw_args['cl_ids'] = self.project_manager.get_classifier_ids()
kw_args['cl_id'] = cl_id
if(self.project_manager.get_classifier_finished(cl_id)):
cv_results, avg_results = pm.get_classifier_result(cl_id)
kw_args['cv_results'] = cv_results
kw_args['avg_results'] = avg_results
kw_args['cl_settings'] = pm.get_classifier_settings(cl_id)
kw_args['cl_names'] = self.CL_NAMES
roc_f = pm.get_roc_f(cl_id)
if(roc_f and os.path.exists(roc_f)):
kw_args['roc_url'] = '%s%s%s/%s' % (self.root_url, self.mm_url,
'roc', cl_id)
else:
kw_args['roc_url'] = None
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)
def new(self, project_id=None, fasta_file=None, sequence_type=None,
use_reference=None, taxon_domain=None, taxon=None):
self.fetch_session_data()
smi = 1
kw_args = self.get_template_args(smi)
# does this stay the same over time???
# taxon domain and corresponding uniprot ancestor numbers
kw_args['taxon_domains'] = self.TAXON_DOMAINS
error_msg = None
# start a new project
if((fasta_file and sequence_type) and project_id):
# taxon domain is not used...
taxon_id = None
if not(use_reference is None):
taxon_id = int(taxon)
if(fasta_file.file is None):
error_msg = 'No fasta file provided'
elif(len(project_id) < 4):
error_msg = 'Project id should be at least 4 characters long'
elif(' ' in project_id):
error_msg = 'Spaces are not allowed in the project id'
elif not(re.match('^[A-Za-z0-9_-]*$', project_id)):
error_msg = 'Only characters, digits, dashes, and ' +\
'underscores are allowed in a project id'
else:
try:
# initiate new project
error_msg = self.project_manager.start_new_project(
project_id, fasta_file, sequence_type,
reference_taxon=taxon_id)
except:
print(traceback.format_exc())
error_msg = 'Error creating new project'
if(error_msg == ''):
# store project id in session
cherrypy.session[self.SESSION_PROJECT_KEY] = project_id
# redirect to project list page
url = self.get_url(0)
raise cherrypy.HTTPRedirect(url)
else:
kw_args['msg'] = error_msg
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)
def list(self):
self.fetch_session_data()
smi = 0
projects = self.project_manager.get_projects()
kw_args = self.get_template_args(smi)
kw_args['projects'] = projects
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)
def calculate(self, featcat_id=None):
smi = 1
self.fetch_session_data()
if self.project_id is None:
return self.no_project_selected()
pm = self.project_manager
fe = pm.get_feature_extraction()
kw_args = self.get_template_args(smi)
kw_args["fe"] = fe
if not (featcat_id is None):
proteins = fe.protein_data_set.proteins
featcat = fe.PROTEIN_FEATURE_CATEGORIES[featcat_id.split("_")[0]]
missing_data = []
# check if data required for feature calculation is available
for get_data_func, all_objects in featcat.required_data:
name = " ".join(get_data_func.__name__.split("_")[1:])
if all_objects:
if not (all([get_data_func(p) for p in proteins])):
missing_data.append(name)
else:
if not (any([get_data_func(p) for p in proteins])):
missing_data.append(name)
if len(missing_data) > 0:
kw_args["msg"] = "<p>Required data is missing: %s</p>" % (", ".join(missing_data))
# check if this feature category is already in the feature matrix
elif featcat_id in fe.available_protein_featcat_ids():
kw_args["msg"] = "This feature category has " + "already been calculated"
else:
# put job in queue
pm.run_feature_extraction([featcat_id])
# redirect to feature list page
raise cherrypy.HTTPRedirect(self.get_url(0))
kw_args["aaindex_scale_ids"] = sequtil.aaindex_scale_ids
kw_args["pseaac_scale_ids"] = sequtil.pseaac_scale_ids
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)
def show_feature_data(self, smi):
self.fetch_session_data()
if self.project_id is None:
return self.no_project_selected()
kw_args = self.get_template_args(smi)
kw_args["fe"] = self.project_manager.get_feature_extraction()
# kw_args['show_filter'] = self.project_manager.get_feat_calc_status()
kw_args["show_filter"] = True
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)
def list(self):
smi = 0
self.fetch_session_data()
if(self.project_id is None):
return self.no_project_selected()
pm = self.project_manager
cl_ids = pm.get_classifier_ids()
kw_args = self.get_template_args(smi)
kw_args['cl_ids'] = cl_ids
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)
def list(self):
smi = 0
self.fetch_session_data()
if self.project_id is None:
return self.no_project_selected()
pm = self.project_manager
fe = pm.get_feature_extraction()
kw_args = self.get_template_args(smi)
kw_args["fe"] = fe
kw_args["featcats"] = fe.PROTEIN_FEATURE_CATEGORIES
kw_args["feat_status"] = pm.get_feat_calc_status()
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)
def load_example(self, example_number):
self.fetch_session_data()
smi = 1
try:
example_number = int(example_number)
except ValueError:
example_number = -1
if(example_number < 0 or example_number >= len(self.EXAMPLES)):
kw_args = self.get_template_args(smi)
template_f = 'no_such_example.html'
return spiceweb.get_template(template_f, **kw_args)
pm = self.project_manager
(pid, seq_f, seq_type, labeling_f) = self.EXAMPLES[example_number]
root_d = spiceweb.spiceweb_dir
seq_f = os.path.join(root_d, self.EXAMPLE_DIR, pid, seq_f)
labeling_f = os.path.join(root_d, self.EXAMPLE_DIR, pid, labeling_f)
error_msg = pm.start_example_project(pid, seq_f, seq_type, labeling_f)
if(error_msg == ''):
# store project id in session
cherrypy.session[self.SESSION_PROJECT_KEY] = pid
# redirect to project list page
url = self.get_url(0)
raise cherrypy.HTTPRedirect(url)
else:
print
print 'This should not happen...'
print error_msg
url = self.get_url(0)
raise cherrypy.HTTPRedirect(url)
def new(self, cl_type=None, n_fold_cv=None, labeling_name=None,
class_ids=None, feat_ids=None):
smi = 1
self.fetch_session_data()
if(self.project_id is None):
return self.no_project_selected()
error_msg = None
# start classification job if required arguments from form are there
if not(cl_type is None or n_fold_cv is None):
pm = self.project_manager
# check the number of class ids
if(len(class_ids) < 2):
error_msg = 'At least two classes should be provided.'
elif(len(feat_ids) < 1):
error_msg = 'No features selected.'
else:
pm.run_classification(cl_type, n_fold_cv, labeling_name,
class_ids, feat_ids)
# redirect to classifier list page
raise cherrypy.HTTPRedirect(self.get_url(0))
# show form otherwise
else:
kw_args = self.get_template_args(smi)
kw_args['fe'] = self.project_manager.get_feature_extraction()
kw_args['show_filter'] = True
kw_args['error_msg'] = error_msg
kw_args['cl_ids'] = self.project_manager.get_classifier_ids()
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)
def details(self, project_id, data_type=None, data_name=None,
data_file=None):
self.fetch_session_data()
smi = 2
# first check if the provided project_id excists
existing_projects = [p[0] for p in self.project_manager.get_projects()]
if not(project_id in existing_projects):
# return message that this project does not exist
kw_args = self.get_template_args(smi)
template_f = 'no_such_project.html'
return spiceweb.get_template(template_f, **kw_args)
# store project id in session
cherrypy.session[self.SESSION_PROJECT_KEY] = project_id
# reset the session data, using the new project id
self.fetch_session_data()
msg_lab = ''
msg_seq = ''
# in case of a data file upload
if((data_type and data_name) and data_file):
pm = self.project_manager
# the upload labeling case
if(data_type == 'labeling'):
# check labeling input data
if(data_file.file is None):
msg_lab = 'No labeling file provided'
elif(' ' in data_name):
msg_lab = 'Spaces are not allowed in the project id'
elif not(re.match('^[A-Za-z0-9_-]*$', data_name)):
msg_lab = 'Only characters, digits, dashes, and ' +\
'underscores are allowed in a project id'
# if no incorrect input data
else:
# try to add the labeling, storing errors in msg_lab
try:
msg_lab = pm.add_labeling(data_name, data_file.file)
except Exception:
print(traceback.format_exc())
msg_lab = 'Error adding labeling'
# chop labeling message to reasonable size
if(len(msg_lab) > 100):
msg_lab = msg_lab[:100] + '...'
# the upload sequence data case
elif(data_type == 'data_source'):
# check sequence input data
if(data_file.file == None):
msg_seq = 'No file provided.'
# if no incorrect input data
else:
# try to add sequence data
try:
msg_seq = pm.add_data_source(data_name, data_file.file)
except Exception:
msg_seq = 'Error adding sequence data.'
if(msg_seq[:13] == 'Error in data'):
msg_seq = msg_seq + '<br /><br />NOTE:<ul><li>Secundary structure sequences should consist of the letters C, H, and E (same as output psipred)</li><li>Solvent accessibility sequences should consist of the letters B (buried), and E (exposed)</li></ul>'
fe = self.project_manager.get_feature_extraction()
kw_args = self.get_template_args(smi)
kw_args['fe'] = fe
kw_args['data_sources'] = ['prot_seq', 'orf_seq', 'ss_seq', 'sa_seq']
kw_args['msg_lab'] = msg_lab
kw_args['msg_seq'] = msg_seq
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)
def no_project_selected(self):
kw_args = self.get_template_args(0)
template_f = "no_project_selected.html"
return spiceweb.get_template(template_f, **kw_args)
def no_such_classifier(self):
kw_args = self.get_template_args(0)
template_f = 'no_such_classifier.html'
return spiceweb.get_template(template_f, **kw_args)
def run(self, cl_id, data_set=None):
smi = 3
self.fetch_session_data()
pm = self.project_manager
fe = pm.get_feature_extraction()
if(self.project_id is None):
return self.no_project_selected()
if not(cl_id in pm.get_classifier_ids()):
return self.no_such_classifier()
kw_args = self.get_template_args(smi)
kw_args['cl_id'] = cl_id
kw_args['cl_ids'] = pm.get_classifier_ids()
if(self.project_manager.get_classifier_finished(cl_id)):
if not(data_set is None):
# required feature categories for running classifier cl_id
settings_dict = pm.get_classifier_settings(cl_id)
feat_ids = settings_dict['feature_names']
feat_cats = set([f.split('_')[0] for f in feat_ids])
# required sequence data for calculating the feature categories
required_seq_data = set()
for fc in feat_cats:
for ds in fe.PROTEIN_FEATURE_CATEGORIES[fc].required_data:
required_seq_data.add(ds)
# SWITCH TO OTHER PROJECT FOR CHECKING SEQUENCE AVAILABILITY
prev_proj = pm.project_id
pm.set_project(data_set)
data_set_fe = pm.get_feature_extraction()
data_set_proteins = data_set_fe.protein_data_set.proteins
# SWITCH BACK TO CURRENT PROJECT
pm.set_project(prev_proj)
# check if required data is available for data set
missing_data = set()
for get_data_func, all_objects in required_seq_data:
name = ' '.join(get_data_func.__name__.split('_')[1:])
print name
print
if(all_objects):
if not(all([get_data_func(p) for p in data_set_proteins])):
missing_data.add(name)
else:
if not(any([get_data_func(p) for p in data_set_proteins])):
missing_data.add(name)
if(len(missing_data) > 0):
# send error msg to template
kw_args['msg'] = '<p>The features that are required for running this classifier can not be calculated for the %s project, because the following sequence data is not available in this project:</p><ul>' % (data_set)
for item in sorted(missing_data):
kw_args['msg'] += '<li>%ss</li>' % (item)
kw_args['msg'] += '</ul>'
else:
# run classifier on data set
pm.run_classify(cl_id, data_set)
# redirect to classifier run page
raise cherrypy.HTTPRedirect(self.get_url(3) + '/' + cl_id)
# fetch all classification data for this classifier
all_data_sets = [p[0] for p in pm.get_projects()]
classification_status = pm.parse_classify_job_files(cl_id)
classification_busy = []
for key in classification_status.keys():
classification_busy.extend(classification_status[key])
classification_unavailable = sorted(set(all_data_sets) -
set(classification_busy))
# forward this data to the template:
# to inform the template if classifier construction has finished
kw_args['classifier_f'] = pm.get_classifier_f(cl_id)
# for the drop-down list of data sets without classification
kw_args['data_sets'] = classification_unavailable
# for the results and status tables
kw_args['classification_status'] = classification_status
template_f = self.get_template_f(smi)
return spiceweb.get_template(template_f, **kw_args)