Skip to content

Clean up the voting system and uncouple the classification. #48

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
bjonnh wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Clean up the voting system and uncouple the classification. #48

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions Classifier/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@

6,721 changes: 6,720 additions & 1 deletion Classifier/dict/index_v1.json
View file
Open in desktop

Large diffs are not rendered by default.

165 changes: 69 additions & 96 deletions Classifier/prediction_voting.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,101 +1,74 @@

import itertools
import numpy as np

def vote_classification(n_path,
n_class,
n_super,
pred_class,
pred_super,
path_from_class,
path_from_superclass, isglycoside, ontology_dictionary):
class_result = []
superclass_result = []
pathway_result = []

index = ontology_dictionary

index_class = list(index['Class'].keys())
index_superclass = list(index['Superclass'].keys())
index_pathway = list(index['Pathway'].keys())

path_for_vote = n_path+path_from_class+path_from_superclass
path = list(set([ k for k in path_for_vote if path_for_vote.count(k) ==3]))

if path == []:
path = list(set([ k for k in path_for_vote if path_for_vote.count(k) ==2]))
if len(path)>1:
path = list(set([ k for k in path_for_vote if path_for_vote.count(k) ==2]))
if path == []:
for w in n_path:
pathway_result.append(index_pathway[w])
return pathway_result,superclass_result,class_result,isglycoside


else: #path != []
if set(n_path) & set(path) != set():
if set(path) & set(path_from_superclass) != set():
n_super = [ l for l in n_super if set(path)& set(index['Super_hierarchy'][str(l)]['Pathway']) != set()]
if n_super == []:
n_class = [ m for m in n_class if set(path) & set(index['Class_hierarchy'][str(m)]['Pathway']) != set() ]
n_super = [index['Class_hierarchy'][str(n)]['Superclass'] for n in n_class]
n_super = list(set(itertools.chain.from_iterable(n_super)))

elif len(n_super) > 1: #super != []
n_class = [ u for u in n_class if set(path) & set(index['Class_hierarchy'][str(u)]['Pathway']) != set() ]
if n_class != []:
n_super = [index['Class_hierarchy'][str(v)]['Superclass'] for v in n_class]
n_path = [index['Class_hierarchy'][str(v)]['Pathway'] for v in n_class]
n_path = list(set(itertools.chain.from_iterable(n_path)))
n_super = list(set(itertools.chain.from_iterable(n_super)))

elif len(path)==1:
n_super = [np.argmax(pred_super)]
n_class = [ m for m in [np.argmax(pred_class)] if set(n_super) & set(index['Class_hierarchy'][str(m)]['Superclass']) != set() ]


else:
n_class = [ o for o in n_class if set(n_super) & set(index['Class_hierarchy'][str(o)]['Superclass'])!=set() ]
if n_class == []:
n_class = [ m for m in [np.argmax(pred_class)] if set(n_super) & set(index['Class_hierarchy'][str(m)]['Superclass']) != set() ]
else:
n_class = [ p for p in n_class if set(path) & set(index['Class_hierarchy'][str(p)]['Pathway']) !=set() ]
n_super = [index['Class_hierarchy'][str(q)]['Superclass'] for q in n_class]

n_super = list(set(itertools.chain.from_iterable(n_super)))

# TODO this can be vastly simplified again


def vote_classification(n_path, n_class, n_super,
pred_class, pred_super,
path_from_class, path_from_superclass,
index):
"""
Classify the obtained classes

@param n_path: predicted pathways above noise
@param n_class: predicted classes above noise
@param n_super: predicted superclasses above noise
@param pred_class: predicted classes (all)
@param pred_super: predicted superclasses (all)
@param path_from_class: pathways extracted from classes
@param path_from_superclass: pathways extracted from superclasses
@param index: the "ontology"
@return: pathway_result, superclass_result, class_result => the classified results ids
"""

pathways_for_vote = list(n_path) + list(path_from_class) + list(path_from_superclass)

# Select pathways with at least 3 counts

path = {k for k in pathways_for_vote if pathways_for_vote.count(k) == 3}

if not path: # if path is empty, we select pathways with only 2 counts
path = {k for k in pathways_for_vote if pathways_for_vote.count(k) == 2}

if not path: # if path is still empty
# Add all the pathways we already know about, the rest is empty
return n_path, [], []

if path & set(n_path): # we have at least some of n_path in our path
if path & set(path_from_superclass): # we have at least some of the path from superclass in path
n_super = [i for i in n_super if path & set(index['Super_hierarchy'][str(i)]['Pathway'])]
n_class = [i for i in n_class if path & set(index['Class_hierarchy'][str(i)]['Pathway'])]

if not n_super: # n_super is empty
n_super = list({index['Class_hierarchy'][str(n)]['Superclass'] for n in n_class})

elif len(n_super) > 1: # super != []
if n_class: # n_class is not empty
n_super = list({j for i in n_class for j in index['Class_hierarchy'][str(i)]['Superclass']})
n_path = list({j for i in n_class for j in index['Class_hierarchy'][str(i)]['Pathway']})
elif len(path) == 1:
n_super = [np.argmax(pred_super)]
best_candidate_class_index = np.argmax(pred_class)
if set(n_super) & set(index['Class_hierarchy'][str(best_candidate_class_index)]['Superclass']):
n_class = [best_candidate_class_index]

else: # we have only one n_super
# now our classes are only the classes from our best superclass
n_class = [i for i in n_class if set(n_super) & set(index['Class_hierarchy'][str(i)]['Superclass'])]
if not n_class: # n_class is empty, we take the predicted class with the highest score
best_candidate_class_index = np.argmax(pred_class)
if set(n_super) & set(index['Class_hierarchy'][str(best_candidate_class_index)]['Superclass']):
n_class = [best_candidate_class_index]

else:
n_super = [ l for l in n_super if set(path) & set(index['Super_hierarchy'][str(l)]['Pathway']) != set()]
if n_super == []:
n_class = [ m for m in n_class if set(path) & set(index['Class_hierarchy'][str(m)]['Pathway']) != set()]
n_super = [index['Class_hierarchy'][str(n)]['Superclass'] for n in n_class]
n_path = [index['Class_hierarchy'][str(v)]['Pathway'] for v in n_class]
n_path = list(set(itertools.chain.from_iterable(n_path)))
n_super = list(set(itertools.chain.from_iterable(n_super)))


elif len(n_super) > 1: #super != []
n_class = [ u for u in n_class if set(path) & set(index['Class_hierarchy'][str(u)]['Pathway']) != set()]
n_super = [index['Class_hierarchy'][str(v)]['Superclass'] for v in n_class]
n_path = [index['Class_hierarchy'][str(v)]['Pathway'] for v in n_class]
n_path = list(set(itertools.chain.from_iterable(n_path)))
n_super = list(set(itertools.chain.from_iterable(n_super)))


else:
n_class = [ o for o in n_class if set(path) & set(index['Class_hierarchy'][str(o)]['Pathway']) != set() ]
n_super = [index['Class_hierarchy'][str(v)]['Superclass'] for v in n_class]
n_path = [index['Class_hierarchy'][str(v)]['Pathway'] for v in n_class]
n_path = list(set(itertools.chain.from_iterable(n_path)))
n_super = list(set(itertools.chain.from_iterable(n_super)))

for r in path:
pathway_result.append(index_pathway[r])
for s in n_super:
superclass_result.append(index_superclass[s])
for t in n_class:
class_result.append(index_class[t])

return pathway_result,superclass_result,class_result,isglycoside #three class results and glycoside checker result (True/False)

n_class = [p for p in n_class if path & set(index['Class_hierarchy'][str(p)]['Pathway'])]
n_super = list({index['Class_hierarchy'][str(q)]['Superclass'] for q in n_class})

else:
# Select all the classes that are part of our selected pathways
n_class = [m for m in n_class if set(path) & set(index['Class_hierarchy'][str(m)]['Pathway'])]
n_super = list({index['Class_hierarchy'][str(n)]['Superclass'] for n in n_class})
n_path = list({index['Class_hierarchy'][str(v)]['Pathway'] for v in n_class})

return path, n_super, n_class # We have to check if we want path or n_path here!
Loading