Perform sliding window analysis on count data in a 2d array

gistfile1.txt

## This code was entirely generated using OpenAIs Code Interpreter and prompting for the type of analysis that was desired

import pandas as pd

# Load the data
data = pd.read_csv('/mnt/data/example_rs_fish.csv')

# Calculate various quantiles
quantiles = data['intensity'].quantile([0.10, 0.25, 0.50, 0.75])
quantiles

# Filter the data based on the 10th percentile threshold
threshold = quantiles[0.10]
filtered_data = data[data['intensity'] > threshold]

# Check the shape of the filtered data
filtered_data.shape

# Binning the data into 10x10 windows
filtered_data['x_bin'] = (filtered_data['x'] // window_size).astype(int)
filtered_data['y_bin'] = (filtered_data['y'] // window_size).astype(int)

# Group by the binned x and y, and calculate the spot count and total intensity
grouped_data = filtered_data.groupby(['x_bin', 'y_bin']).agg(
    spot_count=('intensity', 'count'),
    total_intensity=('intensity', 'sum')
).reset_index()

# Adjust the bin numbers back to pixel numbers
grouped_data['window_x_start'] = grouped_data['x_bin'] * window_size
grouped_data['window_y_start'] = grouped_data['y_bin'] * window_size

# Keep only the columns we need and sort them for readability
grouped_data = grouped_data[['window_x_start', 'window_y_start', 'spot_count', 'total_intensity']]

grouped_data

# Set up the figure and axes for the heatmaps
fig, axs = plt.subplots(2, 1, figsize=(10, 20))

# Create a heatmap for the spot count
spot_count_pivot = grouped_data.pivot('window_y_start', 'window_x_start', 'spot_count')
sns.heatmap(spot_count_pivot, ax=axs[0], cmap='viridis', cbar_kws={'label': 'Spot Count'})
axs[0].set_title('Heatmap of Spot Count')

# Create a heatmap for the total intensity
intensity_pivot = grouped_data.pivot('window_y_start', 'window_x_start', 'total_intensity')
sns.heatmap(intensity_pivot, ax=axs[1], cmap='viridis', cbar_kws={'label': 'Total Intensity'})
axs[1].set_title('Heatmap of Total Intensity')

# Reverse the y-axis to match the image orientation
axs[0].invert_yaxis()
axs[1].invert_yaxis()

plt.tight_layout()
plt.show()

# Create a scatter plot to check for correlation between spot count and total intensity
sns.jointplot(data=grouped_data, x='spot_count', y='total_intensity', kind='reg', color='b')
plt.xlabel('Spot Count')
plt.ylabel('Total Intensity')
plt.title('Correlation between Spot Count and Total Intensity', pad=90)
plt.show()