data['odor_type'] = input("Enter Odor Type: ")
data['seed'] = np.random.randint(0,1000000000)
np.random.seed(data['seed'])
if data['odor_type'] == 'reference':
data['odor_vector'] = data['reference_odor']
print(np.dot(data['odor_vector'],data['reference_odor'])/np.linalg.norm(data['reference_odor'])/np.linalg.norm(data['odor_vector']))
data['odor_angle'] = np.rad2deg(np.arccos(np.clip(np.dot(data['odor_vector'],data['reference_odor'])/np.linalg.norm(data['reference_odor'])/np.linalg.norm(data['odor_vector']),-1,1)))
elif data['odor_type'] == 'no_odor':
data['odor_vector'] = np.zeros(data['dim_odorspace'])
data['odor_angle'] = 0
elif data['odor_type'] == 'random':
data['odor_vector'] = pt.generateUniform(data['odor_concentration'],dimension=data['dim_odorspace'])
data['odor_angle'] = np.rad2deg(np.arccos(np.dot(data['odor_vector'],data['reference_odor'])/np.linalg.norm(data['reference_odor'])/np.linalg.norm(data['odor_vector'])))
elif data['dim_odorspace'] == 2:
data['odor_type'] = int(data['odor_type'])
if np.random.normal() > 0:
theta = np.deg2rad(data['odor_type'])
else:
theta = -np.deg2rad(data['odor_type'])
c, s = np.cos(theta), np.sin(theta)
R = np.array(((c,-s), (s, c)))
data['odor_vector'] = np.matmul(R,data['reference_odor'].reshape(-1,1)).flatten()
data['odor_angle'] = np.rad2deg(np.arccos(np.dot(data['odor_vector'],data['reference_odor'])/np.linalg.norm(data['reference_odor'])/np.linalg.norm(data['odor_vector'])))
elif data['dim_odorspace'] == 3:
data['odor_type'] = int(data['odor_type'])
randVec = pt.generateUniform(1,dimension=data['dim_odorspace'])
crossVec = np.cross(data['reference_odor'],randVec)
def generate_orn(orn_number, duration, resolution, odorVec, odorStart,
odorEnd): # Function to generate single ORN Trace
baseline = locust['baseline_firing'] * (
1 + 0.2 * np.random.normal()
) / locust['peak_firing'] # Baseline Firing Rate Ratio
trace = baseline * np.ones(
int(duration /
resolution)) # Set Baseline activity for the Protocol Duration
rec_field = pt.generateUniform(
1, odor['dim_odorspace'],
seed=int(locust['rec_seeds']
[orn_number])) # Receptive Field of ORNs in Odor Space
latency = locust['latency'][
orn_number] # Latency of Response to Odor Presentation
t_rise = locust['t_rise'][orn_number] # Time to Rise to Peak
t_fall = locust['t_fall'][orn_number] # Response Decay Time
tuning = locust['tuning'][orn_number] / 2 # Odor Tuning-width / Sensitivity
def sigmoid(x, a1=locust['a1'], a2=locust['a2']): # Sigmoid for Response
return 1 / (1 + np.exp(-a1 * (x - a2)))
odorMag = 0.85 * np.linalg.norm(odorVec) # Odor Concentration
cosSim = np.dot(odorVec, rec_field) / (
np.linalg.norm(odorVec) * np.linalg.norm(rec_field)
) # Cosine Similarity wrt Odor
if np.arccos(cosSim) < np.deg2rad(
locust['inh_threshold']): # Minimum Response Threshhold
res_strength = (1 - baseline) * sigmoid(
odorMag * np.cos(np.arccos(cosSim) / 2)**tuning)
else:
res_strength = -baseline * np.linalg.norm(odorVec)
if locust['f_sharp'][orn_number]:
# Generate Sharp Trace
rise = np.arange(0, t_rise / 2, resolution)
rise = baseline + res_strength * 2 * np.exp(
1) / t_rise * rise * np.exp(-2 * rise / t_rise)
riseStartIndex = int((odorStart + latency) / resolution)
riseEndIndex = riseStartIndex + rise.shape[0]
trace[riseStartIndex:riseEndIndex] = rise
peak = rise[-1]
fall = np.linspace(0, duration - riseEndIndex * resolution,
trace.shape[0] - riseEndIndex)
fall = (peak - baseline) * np.exp(-fall / t_fall) + baseline
fallStartIndex = riseEndIndex
trace[fallStartIndex:] = fall
else:
# Generate Broad Trace
rise = np.arange(0, t_rise, resolution)
rise = baseline + res_strength * np.exp(1) / t_rise * rise * np.exp(
-rise / t_rise)
riseStartIndex = int((odorStart + latency) / resolution)
riseEndIndex = int((odorStart + latency) / resolution) + rise.shape[0]
trace[riseStartIndex:riseEndIndex] = rise
peak_1 = rise[-1]
adaptation_rate = locust['adaptation_extent'][
orn_number] # Amplitude of Adaptation-related Decay
t_adaptation = locust['t_adaptation'][
orn_number] # Odor Adaptation Time
adaptation = np.arange(0, (int(odorEnd / resolution) - riseEndIndex) *
resolution, resolution)
adaptation = (peak_1 -
(adaptation_rate * res_strength + baseline)) * np.exp(
-adaptation / t_adaptation) + (
adaptation_rate * res_strength + baseline)
adaptationStartIndex = riseEndIndex
adaptationEndIndex = adaptationStartIndex + adaptation.shape[0]
trace[adaptationStartIndex:adaptationEndIndex] = adaptation
peak_2 = adaptation[-1]
fall = np.arange(0, (trace.shape[0] - adaptationEndIndex) * resolution,
resolution)
fall = (peak_2 - baseline) * np.exp(-fall / t_fall) + baseline
fallStartIndex = adaptationEndIndex
trace[fallStartIndex:] = fall
trace = trace * locust['peak_firing'] * 5 # Scale to Peak Firing Rate
return trace