natural scene decoding ophys¶

This is an example of how to decode natural images from calcium imaging traces in V1

First, let’s download an experiment from the Allen Institute Brain Observatory

from allensdk.core.brain_observatory_cache import BrainObservatoryCache

boc = BrainObservatoryCache()

OEID = 541206592
nwb_dataset = boc.get_ophys_experiment_data(OEID)

Next, we’ll load the dF/F traces and put them in a DataFrame

timestamps, dff = nwb_dataset.get_dff_traces()
neuron_ids = nwb_dataset.get_cell_specimen_ids()

import pandas as pd
traces = pd.DataFrame(
    dff.T,
    columns=neuron_ids,
    index=timestamps,
)

print(traces.head())

Output::

541510267 541510270 541510307 541510405 588381938 541510410 … 541509981 541509952 541510950 541511172 541509957 541511118

10.30338 0.219740 0.151908 0.188632 0.668425 0.081433 0.183986 … 0.226459 0.065945 0.079813 0.084233 0.360267 0.156850 10.33655 0.167939 0.142997 0.178289 0.479256 0.089889 0.073484 … 0.152407 0.127615 0.151770 0.090351 0.308192 0.108658 10.36972 0.136697 0.068048 0.163631 0.533209 0.023668 0.135786 … 0.059235 0.070546 0.114366 0.106841 0.295819 0.033781 10.40289 0.157216 0.105795 0.089224 0.456814 0.004756 0.054532 … 0.138155 0.025575 0.073771 0.085469 0.329818 0.098048 10.43606 0.130490 0.097038 0.099539 0.381433 0.105907 0.050581 … 0.116505 0.037647 0.092018 0.100030 0.487283 0.066526

[5 rows x 154 columns]

Next, we’ll load stim_table

stim_table = nwb_dataset.get_stimulus_table('natural_scenes')
print(stim_table.head())

Output::: frame start end

0 92 16126 16133 1 27 16134 16141 2 52 16141 16148 3 37 16149 16156 4 103 16156 16163

The stim_table lists stimulus times in terms of the start and end frames of the calcium traces, but we need start times and durations for neuroglia, so we’ll need to reshape

stim_table['time'] = timestamps[stim_table['start']]
stim_table['duration'] = timestamps[stim_table['end']+1] - stim_table['time']
stim_table.rename(columns={'frame':'image_id'},inplace=True)

print(stim_table.head())

Output::: image_id start end time duration

0 92 16126 16133 545.22975 0.26537 1 27 16134 16141 545.49512 0.26538 2 52 16141 16148 545.72733 0.26537 3 37 16149 16156 545.99270 0.26538 4 103 16156 16163 546.22491 0.26538

Reduce the traces to responses

import numpy as np
from neuroglia.epoch import EpochTraceReducer

reducer = EpochTraceReducer(traces,func=np.mean)
X = reducer.fit_transform(stim_table[['time','duration']])
y = stim_table['image_id']

Do some dimensionality reduction on the responses

from sklearn.decomposition import PCA
pca = PCA()
X_reduced = pca.fit_transform(X)

Plot the first two Principal Components

import matplotlib.pyplot as plt
plt.scatter(X_reduced[:,0],X_reduced[:,1],c=y,cmap='hsv')

Creating training and test sets

from sklearn.model_selection import train_test_split

X = stim_table[['time','duration']]
y = stim_table['image_id']

chance = 1.0 / len(y.unique())

X_train, X_test, y_train, y_test = train_test_split(X,y,test_size=0.33)

Create a pipeline to decode image

from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis

calcium_traces_pipeline = Pipeline([
    ('reducer', EpochTraceReducer(traces,func=np.mean)),
    ('rescale', StandardScaler()),
    ('classify', LinearDiscriminantAnalysis()),
])

calcium_traces_pipeline.fit(X_train,y_train)
calcium_traces_score = calcium_traces_pipeline.score(X_test,y_test)

print("Calcium Traces Score: {:0.2f} (chance: {:0.3f})".format(calcium_traces_score,chance))

Output::: Calcium Traces Score: 0.05 (chance: 0.01)

Extract calcium events

from neuroglia.calcium import CalciumDeconvolver

deconvolver = CalciumDeconvolver()
calcium_events = deconvolver.fit_transform(traces)

calcium_events_pipeline = Pipeline([
    ('reducer', EpochTraceReducer(calcium_events,func=np.mean)),
    ('rescale', StandardScaler()),
    ('classify', LinearDiscriminantAnalysis()),
])

calcium_events_pipeline.fit(X_train,y_train)
calcium_events_score = calcium_events_pipeline.score(X_test,y_test)

print("Calcium Events Score: {:0.2f} (chance: {:0.3f})".format(calcium_events_score,chance))

Output::: Calcium Traces Score: 0.18 (chance: 0.01)

Gallery generated by Sphinx-Gallery