Load and correlate a set of 6 traces (3 for cell1, 3 for cell2)

In [33]:
import numpy as np
import pandas as pd
import scipy
import scipy.io as spio
import seaborn as sns
import matplotlib.pyplot as plt
%matplotlib inline
from os import listdir
from os.path import isfile, join

Load data for each cell and genotype into a dataframe

In [62]:
# Load the neuron A control genotype data into a dataframe
dfAc = pd.read_excel('AwithAchrim_ctrl1_20171013.xlsx',sheet_name=0, header=1)
dfAc=dfAc[0:100]
dfAc[0:3]
Out[62]:
fly1 fly2 fly3 fly4 fly5 fly6 fly7
0 0.312961 -0.580163 -0.598497 0.003492 0.385425 -0.837685 -0.647115
1 -0.545628 0.164981 0.590876 0.070686 0.452542 0.503481 -0.561781
2 0.171227 0.266987 0.051796 -0.511666 -0.210294 -0.271616 0.617057
In [61]:
# Get means
dfAcMean = pd.DataFrame({'AcMean':dfAc.mean(axis=1)})
dfAcMean[0:3]
Out[61]:
AcMean
0 -0.280226
1 0.096451
2 0.016213
In [63]:
# Load the neuron B control genotype data into a dataframe
dfBc = pd.read_excel('BwithAchrim_ctrl1_20171013.xlsx',sheet_name=0, header=1)  
dfBc=dfBc[0:100]
In [65]:
# Get means
dfBcMean = pd.DataFrame({'BcMean':dfBc.mean(axis=1)})
dfBcMean[0:3]
Out[65]:
BcMean
0 -0.175105
1 -0.026465
2 0.236116
In [36]:
# Load the cell B experimental (light-sensitive) genotype data
dfBe = pd.read_excel('BwithAchrim_expt_20171013.xlsx',sheet_name=0, header=1)
dfBe=dfBe[0:100]
dfBe[0:1]
Out[36]:
fly1 fly2 fly3 fly4 fly5 fly6
0 0.481751 -0.351541 -0.181905 -0.223429 0.241106 0.336005
In [66]:
Out[66]:
BeMean
0 0.050331
1 0.236128
2 0.346278
In [37]:
# Load the cell A experimental (light-sensitive) genotype data
dfAe = pd.read_excel('AwithAchrim_expt_20171013.xlsx',sheet_name=0, header=1)
dfAe=dfAe[0:100]
dfAe[0:1]
Out[37]:
fly1 fly2 fly3 fly4 fly5 fly6
0 -0.100264 -0.339643 -0.116948 -0.333907 -2.836228 -2.437536
In [68]:
# Get means
dfAeMean = pd.DataFrame({'AeMean':dfAe.mean(axis=1)})
dfAeMean[0:3]
Out[68]:
AeMean
0 -1.027421
1 -1.134583
2 -1.031204

Join the dataframes

In [69]:
# Fly by fly data
frames = [dfBc, dfAc, dfBe, dfAe]
dfAll = pd.concat(frames,axis=1)
dfAll[0:1]
Out[69]:
fly1 fly2 fly3 fly4 fly5 fly6 fly7 fly1 fly2 fly3 ... fly3 fly4 fly5 fly6 fly1 fly2 fly3 fly4 fly5 fly6
0 -0.65938 -0.372606 -0.179531 -0.182738 -0.134835 0.611704 -0.308351 0.312961 -0.580163 -0.598497 ... -0.181905 -0.223429 0.241106 0.336005 -0.100264 -0.339643 -0.116948 -0.333907 -2.836228 -2.437536

1 rows × 26 columns

In [76]:
# Means for each condition
framesMeans = [dfBcMean, dfBeMean, dfAcMean, dfAeMean]
dfMeans = pd.concat(framesMeans,axis=1)
dfMeans[0:3]
Out[76]:
BcMean BeMean AcMean AeMean
0 -0.175105 0.050331 -0.280226 -1.027421
1 -0.026465 0.236128 0.096451 -1.134583
2 0.236116 0.346278 0.016213 -1.031204

Calculate correlations

In [77]:
# Fly by fly
corrMat = dfAll.corr(method='spearman')
In [78]:
# Means only
corrMatMeans = dfMeans.corr(method='spearman')

Make a heatmap of the Spearman Correlations of the Means

In [79]:
mask = np.zeros_like(corrMatMeans)
mask[np.triu_indices_from(mask)] = True
with sns.axes_style("white"):
    ax = sns.heatmap(corrMatMeans, mask=mask, square=True, annot=True).set_title('Correlations Between a Pair of Cells')

What does this mean?

  • Neuron B in genotype c is highly correlated with Neuron B in genotype e.
  • Neuron A in in genotype c is uncorrelated with Neuron A in genotype e. This is not surprising, since flies are experiencing a flash of light which can activate A in genotype e, but not in genotype c.
  • We wanted to know whether activating A corresponds to a change in correlation between A and B, so we want to look at Corr(BcMean,AcMean) vs Corr(BeMean,AeMean). Corr(BcMean,AcMean) = 0.47 and Corr(BeMean,AeMean) = 0.27, so it appears that the average responses of A and B were less correlated when Neuron A was activated.