def generate_intensity():
"""Generates an intensity data file related to some user action.
Clear: clear the old intensity mapping
Cluster: intensity mapping associated with a topic cluster
Word: intensity mapping associated with a particular word
Study: intensity mapping associated with a study cluster"""
clicked_on = request.args.get("options")
if clicked_on == "clear":
intensities_by_location = {}
elif clicked_on == "cluster" or clicked_on == "word":
if clicked_on == "cluster":
cluster = request.args.get("cluster")
word = TermCluster.get_words_in_cluster(cluster)
else:
word = request.args.get("word")
studies = StudyTerm.get_by_word(word)
# Create a dictionary of {pmid: frequency} values
frequencies_by_pmid, max_intensity = organize_frequencies_by_study(studies)
pmids = frequencies_by_pmid.keys()
# Get the activations for the keys of the dictionary
activations = Activation.get_activations_from_studies(pmids)
# Assemble the final dictionary of {location:intensity} values, scaling
# each value as we go
intensities_by_location = scale_frequencies_by_loc(activations, max_intensity, frequencies_by_pmid)
elif clicked_on == "study":
pmid = request.args.get("pmid")
study = Study.get_study_by_pmid(pmid)
# Look for cluster-mate studies
cluster_mates = study.get_cluster_mates()
# Get (location, study count) tuples from db
activations = Activation.get_location_count_from_studies(cluster_mates)
# Scale study counts in preparation for intensity mapping
intensities_by_location = scale_study_counts(activations)
print "Found intensities: ", intensities_by_location
# Assemble the intensity map
intensity_vals = generate_intensity_map(intensities_by_location)
return intensity_vals
def generate_citations(radius=3):
"""Returns a list of text citations associated with some location, word
or topic (cluster)."""
clicked_on = request.args.get("options")
if clicked_on == "location":
x_coord = float(request.args.get("xcoord"))
y_coord = float(request.args.get("ycoord"))
z_coord = float(request.args.get("zcoord"))
pmids = Activation.get_pmids_from_xyz(x_coord, y_coord, z_coord, radius)
elif clicked_on == "word":
word = request.args.get("word")
# Get the pmids for a word
pmids = StudyTerm.get_pmid_by_term(word)
elif clicked_on == "cluster":
cluster = request.args.get("cluster")
# Get the words for a cluster
# Then get the top studies for the words
words = TermCluster.get_words_in_cluster(cluster)
pmids = StudyTerm.get_pmid_by_term(words)
elif clicked_on == "study":
pmid = request.args.get("pmid")
study = Study.get_study_by_pmid(pmid)
# Look for cluster-mate studies
pmids = study.get_cluster_mates()
citations = Study.get_references(pmids)
return jsonify(citations)
def load_studies():
"""Load data from database.txt into Location, Activation, Study tables."""
# Delete all rows in existing tables, so if we need to run this a second time,
# we won't add duplicates
Location.query.delete()
Study.query.delete()
Activation.query.delete()
skip = True
count_studies = 0
# Parse txt file and convert to appropriate data types for seeding
for row in open("seed_data/database.txt"):
# Skip the header of the txt file
if skip:
skip = False
continue
# Stop after the first 5000 rows for now
# if count_studies > 5000:
# break
row = row.rstrip().split("\t")
# Information to go into Study, if applicable:
pmid = int(row[0])
doi = row[1]
title = row[9]
authors = row[10]
year = int(row[11])
journal = row[12].rstrip()
# Information to go into Location, if applicable
x = float(row[2])
y = float(row[3])
z = float(row[4])
space = row[5]
# Check whether PMID is already in Study; if not, add it to db.
study_obj = Study.get_study_by_pmid(pmid)
if study_obj is None:
study_to_add = Study(pmid=pmid, doi=doi, title=title, authors=authors, year=year, journal=journal)
db.session.add(study_to_add)
db.session.commit()
# Check whether xyz is already in Location; if not, add it to db and
# retrieve its location ID (an autoincrementing primary key).
# If xyz already in Location, retrieve its location_id.
location_obj = Location.check_by_xyz_space(x, y, z, space)
if location_obj is None:
location_to_add = Location(x_coord=x, y_coord=y, z_coord=z, space=space)
db.session.add(location_to_add)
db.session.commit()
loc_id = Location.check_by_xyz_space(x, y, z, space).location_id
else:
loc_id = location_obj.location_id
# Add activation to db, using location_id identified/generated above
activation_to_add = Activation(pmid=pmid, location_id=loc_id)
db.session.add(activation_to_add)
db.session.commit()
# Print where we are and increment counter
print "Database.txt seeding row ", count_studies
count_studies += 1
def load_studies():
"""Loads data from database.txt into Location, Activation, Study tables.
File format: PMID \t doi \t x \t y \t z \t space \t peak_id \t table_id
\t table_num \t title \t authors \t year \t journal \t
Source: Neurosynth database.txt file"""
skip = True
count_studies = 0
database = open("seed_data/database.txt")
# Parse txt file and convert to appropriate data types for seeding
for row in database:
# Skip the header of the txt file
if skip:
skip = False
continue
# Stop after the first 5000 rows for now
# if count_studies > 5000:
# break
row = row.rstrip().split('\t')
# Information to go into Study, if applicable:
pmid = int(row[0])
doi = row[1]
title = row[9]
authors = row[10]
year = int(row[11])
journal = row[12].rstrip()
# Information to go into Location, if applicable
x = float(row[2])
y = float(row[3])
z = float(row[4])
# Check whether PMID is already in Study; if not, add it to db.
study_obj = Study.get_study_by_pmid(pmid)
if study_obj is None:
study_to_add = Study(pmid=pmid, doi=doi, title=title,
authors=authors, year=year, journal=journal)
db.session.add(study_to_add)
db.session.commit()
# Check whether xyz is already in Location; if not, add it to db and
# retrieve its location ID (an autoincrementing primary key).
# If xyz already in Location, get its location_id.
location_obj = Location.check_by_xyz(x, y, z)
if location_obj is None:
location_to_add = Location(x_coord=x, y_coord=y, z_coord=z)
db.session.add(location_to_add)
db.session.commit()
loc_id = Location.check_by_xyz(x, y, z).location_id
else:
loc_id = location_obj.location_id
# Add activation to db, using location_id identified/generated above
activation_to_add = Activation(pmid=pmid, location_id=loc_id)
db.session.add(activation_to_add)
db.session.commit()
# Print where we are and increment counter
print "Database.txt seeding row ", count_studies
count_studies += 1
database.close()
def generate_d3(radius=3):
""" Returns JSON with xyz at the root node.
Test with parameters: 40, -45, -25 (Fusiform face area)
"""
clicked_on = request.args.get("options")
if clicked_on == "location":
x_coord = float(request.args.get("xcoord"))
y_coord = float(request.args.get("ycoord"))
z_coord = float(request.args.get("zcoord"))
pmids = Activation.get_pmids_from_xyz(x_coord, y_coord, z_coord, radius)
scale = 70000
# Get [(wd, freq), ...] and [wd1, wd2] for most frequent words
elif clicked_on == "study":
pmid = request.args.get("pmid")
study = Study.get_study_by_pmid(pmid)
pmids = study.get_cluster_mates()
scale = 30000
terms_for_dict, words = StudyTerm.get_terms_by_pmid(pmids)
# Optional: transform the terms
# Get the top clusters
top_clusters = TermCluster.get_top_clusters(words)
# Get the cluster-word associations
associations = TermCluster.get_word_cluster_pairs(top_clusters, words)
# Make the root node:
root_dict = {"name": "", "children": []}
# Build the terminal nodes (leaves) first using (wd, freq) tuples
# Output: {word: {'name': word, 'size': freq}, word2: ... }
leaves = {}
for (word, freq) in terms_for_dict:
if word not in leaves:
leaves[word] = {"name": word, "size": freq * scale}
else:
leaves[word]["size"] += freq * scale
# Embed the leaves in the clusters:
# Output: {cluster_id: {'name': ID, 'children': [...]}, ... }
clusters = {}
for (cluster_id, word) in associations:
if cluster_id not in clusters:
clusters[cluster_id] = {"name": cluster_id, "children": [leaves[word]]}
else:
clusters[cluster_id]["children"].append(leaves[word])
# Put the clusters in the root dictionary
# Output: {'name': root, children: [{'name': id, 'children': []}, ...]
for cluster in top_clusters:
root_dict["children"].append(clusters[cluster])
return jsonify(root_dict)
