def __init__(self, frontFileName, sideFileName, height, ellipticTable,
bau):
self.frontFigure = Length.Length(frontFileName, bau)
self.frontFigure.input_height(height)
self.sideFigure = Length.Length(sideFileName, bau)
self.sideFigure.input_height(height)
self.ellipticTable = ellipticTable
def generate_model(input_shape_cdr3, num_outputs, filter_size):
features_cdr3 = Input(shape=input_shape_cdr3)
features_quantity = Input(shape=[])
feature_age = Input(batch_shape=[1])
weight = Input(batch_shape=[1])
level = Input(batch_shape=[1])
features_mask = Masking(mask_value=0.0)(features_cdr3)
features_length = Length()(features_mask)
features_abundance = Abundance()(features_quantity)
features_age = BatchExpand()([feature_age, features_abundance])
weights_instance = Multiply()([weight, features_quantity])
num_filters = 2 * num_outputs # Two instances are required for each output
logits_cdr3 = Alignment(num_filters,
filter_size,
penalties_feature=0.0,
penalties_filter=-1.0E16,
length_normalize=False)(features_mask)
logits_cdr3_norm = NormalizeInitializationByAggregation(
1, epsilon=1.0E-5)([logits_cdr3, weights_instance, level])
feature_length_norm = NormalizeInitializationByAggregation(
0, epsilon=1.0E-5)([features_length, weights_instance, level])
logits_length = Dense(num_filters)(feature_length_norm)
logits_length_norm = NormalizeInitializationByAggregation(
1, epsilon=1.0E-5)([logits_length, weights_instance, level])
features_abundance_norm = NormalizeInitializationByAggregation(
0, epsilon=1.0E-5)([features_abundance, weights_instance, level])
logits_abundance = Dense(num_filters)(features_abundance_norm)
logits_abundance_norm = NormalizeInitializationByAggregation(
1, epsilon=1.0E-5)([logits_abundance, weights_instance, level])
features_age_norm = NormalizeInitializationByAggregation(
0, epsilon=1.0E-5)([features_age, weights_instance, level])
logits_age = Dense(num_filters)(features_age_norm)
logits_age_norm = NormalizeInitializationByAggregation(
1, epsilon=1.0E-5)([logits_age, weights_instance, level])
logits = Add()([
logits_cdr3_norm, logits_length_norm, logits_abundance_norm,
logits_age_norm
])
logits_aggregate = Aggregate2Instances()(logits) # Uses to instances
logits_aggregate_norm = NormalizeInitializationByAggregation(
2, epsilon=1.0E-5)([logits_aggregate, weight, level])
logits_flat = FullFlatten()(logits_aggregate_norm)
model = Model(
inputs=[features_cdr3, features_quantity, feature_age, weight, level],
outputs=logits_flat)
return model
class RegistrationForm(FlaskForm):
username =StringField('Usrname',validators=[DataRequired(),Length(min=2, max=20)])
email =StringField('Email',validators=[DataRequired(),Email()])
password =PasswordField('Password',validators=[DataRequired()])
confirm_password =PasswordField('Confirm Password',validators=[DataRequired(),Equalto('password')])
submit =SubmitField('Sign Up')
class LoginForm(FlaskForm):
email =StringField('Email',validators=[DataRequired(),Email()])
password =PasswordField('Password',validators=[DataRequired()])
remember= BoolenField('Remember Me')
submit =SubmitField('Login')
class RegistrationForm(FlaskForm):
username = StringField('Username',
validators=[DataRequired(), Length(min=2, max=20)])
email= StringField('Email',
validators=[DataRequired(), Email()])
password= StringField('Password',
validators=[DataRequired()])
confirm_password= StringField('Confirm Password',
validators=[DataRequired(), EqualTo('password')])
submit = SubmitField('Sign Up')
def generate_model(input_shape_cdr3, num_outputs, filter_size):
features_cdr3 = Input(shape=input_shape_cdr3)
features_quantity = Input(shape=[])
feature_age = Input(batch_shape=[1])
weight = Input(batch_shape=[1])
level = Input(batch_shape=[1])
features_mask = Masking(mask_value=0.0)(features_cdr3)
features_length = Length()(features_mask)
features_abundance = Abundance()(features_quantity)
features_age = BatchExpand()([ feature_age, features_abundance ])
weights_instance = Multiply()([weight, features_quantity])
logits_cdr3 = Conv1D(num_outputs, filter_size)(features_cdr3)
logits_cdr3_mask = MaskCopy(trim_front=filter_size-1)([ logits_cdr3, features_mask ])
logits_cdr3_pool = GlobalPoolWithMask()(logits_cdr3_mask)
logits_cdr3_norm = NormalizeInitializationByAggregation(1, epsilon=1.0E-5)([ logits_cdr3_pool, weights_instance, level ])
feature_length_norm = NormalizeInitializationByAggregation(0, epsilon=1.0E-5)([ features_length, weights_instance, level ])
logits_length = Dense(num_outputs)(feature_length_norm)
logits_length_norm = NormalizeInitializationByAggregation(1, epsilon=1.0E-5)([ logits_length, weights_instance, level ])
features_abundance_norm = NormalizeInitializationByAggregation(0, epsilon=1.0E-5)([ features_abundance, weights_instance, level ])
logits_abundance = Dense(num_outputs)(features_abundance_norm)
logits_abundance_norm = NormalizeInitializationByAggregation(1, epsilon=1.0E-5)([ logits_abundance, weights_instance, level ])
features_age_norm = NormalizeInitializationByAggregation(0, epsilon=1.0E-5)([ features_age, weights_instance, level ])
logits_age = Dense(num_outputs)(features_age_norm)
logits_age_norm = NormalizeInitializationByAggregation(1, epsilon=1.0E-5)([ logits_age, weights_instance, level ])
logits = Add()(
[ logits_cdr3_norm, logits_length_norm, logits_abundance_norm, logits_age_norm ]
)
logits_aggregate = Aggregate()(logits)
logits_aggregate_norm = NormalizeInitializationByAggregation(2, epsilon=1.0E-5)([ logits_aggregate, weight, level ])
logits_flat = FullFlatten()(logits_aggregate_norm)
model = Model(
inputs=[ features_cdr3, features_quantity, feature_age, weight, level ],
outputs=logits_flat
)
return model
def generate_model(input_shape_tra_cdr3, input_shape_tra_vgene,
input_shape_tra_jgene, input_shape_trb_cdr3,
input_shape_trb_vgene, input_shape_trb_jgene, num_outputs):
features_tra_cdr3 = Input(shape=input_shape_tra_cdr3)
features_tra_vgene = Input(shape=input_shape_tra_vgene)
features_tra_jgene = Input(shape=input_shape_tra_jgene)
features_trb_cdr3 = Input(shape=input_shape_trb_cdr3)
features_trb_vgene = Input(shape=input_shape_trb_vgene)
features_trb_jgene = Input(shape=input_shape_trb_jgene)
weights = Input(shape=[])
features_tra_mask = Masking(mask_value=0.0)(features_tra_cdr3)
features_tra_length = Length()(features_tra_mask)
logits_tra_cdr3 = LSTM(num_outputs)(features_tra_mask)
logits_tra_cdr3_norm = NormalizeInitialization(epsilon=0.0)(
[logits_tra_cdr3, weights])
logits_tra_length = Dense(num_outputs)(features_tra_length)
logits_tra_length_norm = NormalizeInitialization(epsilon=0.0)(
[logits_tra_length, weights])
logits_tra_vgene = Dense(num_outputs)(features_tra_vgene)
logits_tra_vgene_norm = NormalizeInitialization(epsilon=0.0)(
[logits_tra_vgene, weights])
logits_tra_jgene = Dense(num_outputs)(features_tra_jgene)
logits_tra_jgene_norm = NormalizeInitialization(epsilon=0.0)(
[logits_tra_jgene, weights])
features_trb_mask = Masking(mask_value=0.0)(features_trb_cdr3)
features_trb_length = Length()(features_trb_mask)
logits_trb_cdr3 = LSTM(num_outputs)(features_trb_mask)
logits_trb_cdr3_norm = NormalizeInitialization(epsilon=0.0)(
[logits_trb_cdr3, weights])
logits_trb_length = Dense(num_outputs)(features_trb_length)
logits_trb_length_norm = NormalizeInitialization(epsilon=0.0)(
[logits_trb_length, weights])
logits_trb_vgene = Dense(num_outputs)(features_trb_vgene)
logits_trb_vgene_norm = NormalizeInitialization(epsilon=0.0)(
[logits_trb_vgene, weights])
logits_trb_jgene = Dense(num_outputs)(features_trb_jgene)
logits_trb_jgene_norm = NormalizeInitialization(epsilon=0.0)(
[logits_trb_jgene, weights])
logits = Add()([
logits_tra_cdr3_norm, logits_tra_length_norm, logits_tra_vgene_norm,
logits_tra_jgene_norm, logits_trb_cdr3_norm, logits_trb_length_norm,
logits_trb_vgene_norm, logits_trb_jgene_norm
])
logits_norm = NormalizeInitialization(epsilon=0.0)([logits, weights])
model = Model(inputs=[
features_tra_cdr3, features_tra_vgene, features_tra_jgene,
features_trb_cdr3, features_trb_vgene, features_trb_jgene, weights
],
outputs=logits_norm)
return model
def main():
container = Container()
#obtain configs in a list format
config = open(sys.argv[1]).read().splitlines()
# obtain the problem file and throw it into a list object
container.shapes = open(sys.argv[2]).read().splitlines()
# Variables that will be used to set the 2d array of material
container.maxWidth = container.shapes[0].split(" ")[0]
container.maxLength = Length.getLength(container.shapes)
#number of shapes in the problem file
container.numShapes = container.shapes[0].split(" ")[1]
# delete the width and number of shapes from the shape list
del container.shapes[0]
# User decided on the Randomized run
if sys.argv[3] == "Random":
# setting up variables using config file
if sys.argv[4] == "newSeed":
for rules in config:
info = rules.split(" ")
if info[0] == "fitness_evaluations":
container.evaluations = info[1]
elif info[0] == "runs":
container.numRuns = info[1]
elif info[0] == "prob_log":
container.prob_log_file = info[1]
elif info[0] == "prob_solution":
container.prob_solution_file = info[1]
elif info[0] == "seed":
container.seed = eval(info[1])
elif sys.argv[4] == "lastSeed":
for rules in config:
info = rules.split(" ")
if info[0] == "fitness_evaluations":
container.evaluations = info[1]
elif info[0] == "runs":
container.numRuns = info[1]
elif info[0] == "prob_log":
container.prob_log_file = info[1]
elif info[0] == "prob_solution":
container.prob_solution_file = info[1]
elif info[0] == "seed":
obtain_seed = open(
container.prob_log_file).read().splitlines(3)
for lines in obtain_seed:
line = lines.split(" ")
if line[0] == "Random":
container.seed = line[3]
break
# Seeds the random function using a saved value that is put into the log file
random.seed(container.seed)
# opening the log file
result_log = open(container.prob_log_file, 'w')
# formatting the result log with Result Log at the top
result_log.write("Result Log \n")
result_log.write("Problem Instance Path = ../%s \n" % sys.argv[2])
result_log.write("Random Seed = %s \n" % container.seed)
result_log.write(
"Parameters used = {'fitness evaluations': %s, 'number of runs': %s, 'problem solution location': '%s'}\n\n"
% (container.evaluations, container.numRuns,
container.prob_solution_file))
# runs through the program as many times as the config files says to
for run in range(1, int(container.numRuns) + 1):
# highest fitness calculation thus far this run
highest_fitness = 0
# Open the current solution file to obtain the fitness value
solution_file = open(
container.prob_solution_file).read().splitlines()
# grabs the solution file's fitness value
solution_fitness = solution_file[1].split(" ")[3]
# Titles each section with Run i, where i is the run number (1-30)
result_log.write("Run " + str(run) + "\n")
# run through the given amount of times given by fitness evaluation
for fitness in range(1, int(container.evaluations) + 1):
# list of solution locations incase it is the best
solution_locations = []
# holders for length of material used
LargestX = 0
SmallestX = 156
# the material sheet being used to cut out shapes
container.materialSheet = [[
0 for x in range(0, int(container.maxWidth))
] for y in range(0, int(container.maxLength))]
# for every shape in the file, choose a position
for shape in container.shapes:
if not shape[0].isdigit():
valid = False
# Keep obtaining a new position until it fits on the material
while not valid:
# generate random position and rotation
x_cord = random.randrange(0,
int(container.maxLength))
y_cord = random.randrange(0,
int(container.maxWidth))
rotation = random.randrange(0, 4)
# Rotate the shape if needed
if rotation != 0:
shape = rotate.rotate_shape(rotation, shape)
# Check whether the shape fits on the material in the current position
valid = shapeManipulation.validPlacement(
container.materialSheet, container.maxLength,
container.maxWidth, x_cord, y_cord, shape)
# if the move was valid and was placed
if valid:
shapeManipulation.placeShape(
container.materialSheet, x_cord, y_cord,
shape)
# store the location in a tuple if it worked
placementLocation = (x_cord, y_cord, rotation)
# append it to the list
solution_locations.append(placementLocation)
# obtains the smallest and largest position in the material array
for i in range(len(container.materialSheet)):
if 1 in container.materialSheet[i]:
if i < SmallestX:
SmallestX = i
elif i > LargestX:
LargestX = i
# Determines the Length of the material used by this iteration
usedLength = ((LargestX - SmallestX) + 1)
#print(usedLength)
current_fitness = fitnessCalc(container.maxLength, usedLength)
if highest_fitness < current_fitness:
highest_fitness = current_fitness
result_log.write(
str(fitness + 1) + " " + str(current_fitness) + "\n")
# If the current solution is the best, replace the info in the file with the new solution
if int(solution_fitness) < highest_fitness:
solution_file = open(prob_solution_file, 'w')
solution_file.write("Solution File\n")
solution_file.write("Fitness Calculation = " +
str(highest_fitness) + "\n\n")
for i in range(len(solution_locations)):
solution_file.write(
str(solution_locations[i])[1:-1] + "\n")
# formatting the result log with a space after each run block
result_log.write("\n")
result_log.close()
def generate_model(input_shape_tra_cdr3, input_shape_tra_vgene,
input_shape_tra_jgene, input_shape_trb_cdr3,
input_shape_trb_vgene, input_shape_trb_jgene, num_outputs):
kmer_size = 4
features_tra_cdr3 = Input(shape=input_shape_tra_cdr3)
features_tra_vgene = Input(shape=input_shape_tra_vgene)
features_tra_jgene = Input(shape=input_shape_tra_jgene)
features_trb_cdr3 = Input(shape=input_shape_trb_cdr3)
features_trb_vgene = Input(shape=input_shape_trb_vgene)
features_trb_jgene = Input(shape=input_shape_trb_jgene)
weights = Input(shape=[])
features_tra_mask = Masking(mask_value=0.0)(features_tra_cdr3)
features_tra_length = Length()(features_tra_mask)
logits_tra_cdr3 = Conv1D(8, kmer_size)(features_tra_cdr3)
logits_tra_cdr3 = Conv1D(num_outputs, kmer_size)(logits_tra_cdr3)
logits_tra_cdr3_mask = MaskCopy(trim_front=2 * kmer_size -
2)([logits_tra_cdr3, features_tra_mask])
logits_tra_cdr3_pool = GlobalPoolWithMask()(logits_tra_cdr3_mask)
logits_tra_cdr3_norm = NormalizeInitialization(epsilon=0.0)(
[logits_tra_cdr3_pool, weights])
logits_tra_length = Dense(num_outputs)(features_tra_length)
logits_tra_length_norm = NormalizeInitialization(epsilon=0.0)(
[logits_tra_length, weights])
logits_tra_vgene = Dense(num_outputs)(features_tra_vgene)
logits_tra_vgene_norm = NormalizeInitialization(epsilon=0.0)(
[logits_tra_vgene, weights])
logits_tra_jgene = Dense(num_outputs)(features_tra_jgene)
logits_tra_jgene_norm = NormalizeInitialization(epsilon=0.0)(
[logits_tra_jgene, weights])
features_trb_mask = Masking(mask_value=0.0)(features_trb_cdr3)
features_trb_length = Length()(features_trb_mask)
logits_trb_cdr3 = Conv1D(8, kmer_size)(features_trb_cdr3)
logits_trb_cdr3 = Conv1D(num_outputs, kmer_size)(logits_trb_cdr3)
logits_trb_cdr3_mask = MaskCopy(trim_front=2 * kmer_size -
2)([logits_trb_cdr3, features_tra_mask])
logits_trb_cdr3_pool = GlobalPoolWithMask()(logits_trb_cdr3_mask)
logits_trb_cdr3_norm = NormalizeInitialization(epsilon=0.0)(
[logits_trb_cdr3_pool, weights])
logits_trb_length = Dense(num_outputs)(features_trb_length)
logits_trb_length_norm = NormalizeInitialization(epsilon=0.0)(
[logits_trb_length, weights])
logits_trb_vgene = Dense(num_outputs)(features_trb_vgene)
logits_trb_vgene_norm = NormalizeInitialization(epsilon=0.0)(
[logits_trb_vgene, weights])
logits_trb_jgene = Dense(num_outputs)(features_trb_jgene)
logits_trb_jgene_norm = NormalizeInitialization(epsilon=0.0)(
[logits_trb_jgene, weights])
logits = Add()([
logits_tra_cdr3_norm, logits_tra_length_norm, logits_tra_vgene_norm,
logits_tra_jgene_norm, logits_trb_cdr3_norm, logits_trb_length_norm,
logits_trb_vgene_norm, logits_trb_jgene_norm
])
logits_norm = NormalizeInitialization(epsilon=0.0)([logits, weights])
model = Model(inputs=[
features_tra_cdr3, features_tra_vgene, features_tra_jgene,
features_trb_cdr3, features_trb_vgene, features_trb_jgene, weights
],
outputs=logits_norm)
return model
def generate_model(input_shape_tra_cdr3, input_shape_tra_vgene,
input_shape_tra_jgene, input_shape_trb_cdr3,
input_shape_trb_vgene, input_shape_trb_jgene, num_outputs):
features_tra_cdr3 = Input(shape=input_shape_tra_cdr3)
features_tra_vgene = Input(shape=input_shape_tra_vgene)
features_tra_jgene = Input(shape=input_shape_tra_jgene)
features_trb_cdr3 = Input(shape=input_shape_trb_cdr3)
features_trb_vgene = Input(shape=input_shape_trb_vgene)
features_trb_jgene = Input(shape=input_shape_trb_jgene)
features_tra_mask = Masking(mask_value=0.0)(features_tra_cdr3)
features_tra_length = Length()(features_tra_mask)
logits_tra_cdr3 = Alignment(num_outputs,
input_shape_tra_cdr3[0],
penalties_feature=-1.0E16,
penalties_filter=0.0,
length_normalize=True)(features_tra_mask)
logits_tra_cdr3_norm = BatchNormalization(momentum=0.5)(logits_tra_cdr3)
logits_tra_length = Dense(num_outputs)(features_tra_length)
logits_tra_length_norm = BatchNormalization(
momentum=0.5)(logits_tra_length)
logits_tra_vgene = Dense(num_outputs)(features_tra_vgene)
logits_tra_vgene_norm = BatchNormalization(momentum=0.5)(logits_tra_vgene)
logits_tra_jgene = Dense(num_outputs)(features_tra_jgene)
logits_tra_jgene_norm = BatchNormalization(momentum=0.5)(logits_tra_jgene)
features_trb_mask = Masking(mask_value=0.0)(features_trb_cdr3)
features_trb_length = Length()(features_trb_mask)
logits_trb_cdr3 = Alignment(num_outputs,
input_shape_trb_cdr3[0],
penalties_feature=-1.0E16,
penalties_filter=0.0,
length_normalize=True)(features_trb_mask)
logits_trb_cdr3_norm = BatchNormalization(momentum=0.5)(logits_trb_cdr3)
logits_trb_length = Dense(num_outputs)(features_trb_length)
logits_trb_length_norm = BatchNormalization(
momentum=0.5)(logits_trb_length)
logits_trb_vgene = Dense(num_outputs)(features_trb_vgene)
logits_trb_vgene_norm = BatchNormalization(momentum=0.5)(logits_trb_vgene)
logits_trb_jgene = Dense(num_outputs)(features_trb_jgene)
logits_trb_jgene_norm = BatchNormalization(momentum=0.5)(logits_trb_jgene)
logits = Add()([
logits_tra_cdr3_norm, logits_tra_length_norm, logits_tra_vgene_norm,
logits_tra_jgene_norm, logits_trb_cdr3_norm, logits_trb_length_norm,
logits_trb_vgene_norm, logits_trb_jgene_norm
])
logits_norm = BatchNormalization(momentum=0.5)(logits)
probabilities = Activation('softmax')(logits_norm)
model = Model(inputs=[
features_tra_cdr3, features_tra_vgene, features_tra_jgene,
features_trb_cdr3, features_trb_vgene, features_trb_jgene
],
outputs=probabilities)
return model
for key, value in results_dict.items():
li = []
li.append(key)
for v in value:
li.append(v)
li.append(1)
writer.writerow(li)
print("Results saved in results.csv file!")
print("Calculating A and C1")
essays_list = kd_reader.read_essays(data_path_train, train_files)
essay_words, essay_pos = kd_reader.gen_word_pos(essays_list)
essay_tokenized = kd_reader.gen_sent(essays_list)
essay_parsed = kd_reader.gen_parse(essay_tokenized, stanford_path)
length_train = Length.count_sent(essay_pos,
essay_tokenized) #no. of sentences a
c_1_count_train = c_1.count_c1(
essay_parsed) # number of subj-verb agreement mistakes in each essay
essays_list = kd_reader.read_essays(data_path_test, test_files)
essay_words, essay_pos = kd_reader.gen_word_pos(essays_list)
essay_tokenized = kd_reader.gen_sent(essays_list)
essay_parsed = kd_reader.gen_parse(essay_tokenized, stanford_path)
length_test = Length.count_sent(essay_pos,
essay_tokenized) #no. of sentences a
c_1_count_test = c_1.count_c1(essay_parsed)
#print(c_1_count_test,c_1_count_train,length_test,length_train)
df_train = pd.DataFrame()
df_train['length_train'] = length_train
df_train['c1_train'] = c_1_count_train
def generate_model(
input_shape_tra_cdr3, input_shape_tra_vgene, input_shape_tra_jgene,
input_shape_trb_cdr3, input_shape_trb_vgene, input_shape_trb_jgene,
num_outputs
):
features_tra_cdr3 = Input(shape=input_shape_tra_cdr3)
features_tra_vgene = Input(shape=input_shape_tra_vgene)
features_tra_jgene = Input(shape=input_shape_tra_jgene)
features_trb_cdr3 = Input(shape=input_shape_trb_cdr3)
features_trb_vgene = Input(shape=input_shape_trb_vgene)
features_trb_jgene = Input(shape=input_shape_trb_jgene)
weights = Input(shape=[])
features_tra_mask = Masking(mask_value=0.0)(features_tra_cdr3)
features_tra_length = Length()(features_tra_mask)
logits_tra_cdr3 = Alignment(num_outputs, input_shape_tra_cdr3[0], penalties_feature=-1.0E16, penalties_filter=0.0, length_normalize=True)(features_tra_mask)
logits_tra_cdr3_norm = WeightedInitNormalization(epsilon=0.0)([ logits_tra_cdr3, weights ])
logits_tra_length = Dense(num_outputs)(features_tra_length)
logits_tra_length_norm = WeightedInitNormalization(epsilon=0.0)([ logits_tra_length, weights ])
logits_tra_vgene = Dense(num_outputs)(features_tra_vgene)
logits_tra_vgene_norm = WeightedInitNormalization(epsilon=0.0)([ logits_tra_vgene, weights ])
logits_tra_jgene = Dense(num_outputs)(features_tra_jgene)
logits_tra_jgene_norm = WeightedInitNormalization(epsilon=0.0)([ logits_tra_jgene, weights ])
features_trb_mask = Masking(mask_value=0.0)(features_trb_cdr3)
features_trb_length = Length()(features_trb_mask)
logits_trb_cdr3 = Alignment(num_outputs, input_shape_trb_cdr3[0], penalties_feature=-1.0E16, penalties_filter=0.0, length_normalize=True)(features_trb_mask)
logits_trb_cdr3_norm = WeightedInitNormalization(epsilon=0.0)([ logits_trb_cdr3, weights ])
logits_trb_length = Dense(num_outputs)(features_trb_length)
logits_trb_length_norm = WeightedInitNormalization(epsilon=0.0)([ logits_trb_length, weights ])
logits_trb_vgene = Dense(num_outputs)(features_trb_vgene)
logits_trb_vgene_norm = WeightedInitNormalization(epsilon=0.0)([ logits_trb_vgene, weights ])
logits_trb_jgene = Dense(num_outputs)(features_trb_jgene)
logits_trb_jgene_norm = WeightedInitNormalization(epsilon=0.0)([ logits_trb_jgene, weights ])
logits = Add()(
[
logits_tra_cdr3_norm, logits_tra_length_norm, logits_tra_vgene_norm, logits_tra_jgene_norm,
logits_trb_cdr3_norm, logits_trb_length_norm, logits_trb_vgene_norm, logits_trb_jgene_norm
]
)
logits_norm = WeightedInitNormalization(epsilon=0.0)([ logits, weights ])
model = Model(
inputs=[
features_tra_cdr3, features_tra_vgene, features_tra_jgene,
features_trb_cdr3, features_trb_vgene, features_trb_jgene,
weights
],
outputs=logits_norm
)
return model
class InputForm(FlaskForm):
symbol = StringField('Symbol',
validators=[DataRequired(), Length(min=2, max=4)])
startdate = StringField('StartDate')
enddate = StringField('EndDate')
submit = SubmitField('Get History')
