Pixel Script

1. Overview
2. Setup CSV File
3. Command-Line Arguments
4. Example Usage
5. Output
6. Troubleshooting

Overview

This command-line script performs operations on pixel data, including k-means clustering and correlation heatmaps. The clustering assigns each pixel event to a specific cluster, and the heatmap helps identify correlations between different channels.

Setup CSV File

The setup.csv file is used to configure how the script operates on data. It specifies which channels from the data should be included in the clustering and heatmap generation steps. Here's how to structure the setup.csv file:

• The first column, labeled channels, should contain the names of the desired channels within the data.
• The second column, labeled cluster, should contain a 1 if the corresponding channel should be included in clustering, and a 0 otherwise.
• The third column, labeled heatmap, should contain a 1 if the corresponding channel should be included in the heatmap, and a 0 otherwise.

Example setup.csv

channels	cluster	heatmap
Intensity_IntegratedIntensity_iron_filtered_c1	1	1
Intensity_IntegratedIntensity_cobalt_filtered_c1	0	1
Intensity_IntegratedIntensity_nuclei_filtered_c1	1	0

This setup would include the Intensity_IntegratedIntensity_iron_filtered_c1 channel in both the clustering and heatmap steps. The Intensity_IntegratedIntensity_cobalt_filtered_c1 channel would only be included in the heatmap, and the Intensity_IntegratedIntensity_nickel_filtered_c1 would only be included in the clustering.

Command-Line Arguments

Required Arguments

• --data_csv: Specifies the path to the data CSV file.

  Linux/Mac: --data_csv="/path/to/data.csv"

  Windows: --data_csv="C:\path\to\data.csv"

• --setup_csv: Specifies the path to the setup CSV file.

  Linux/Mac: --setup_csv="/path/to/setup.csv"

  Windows: --setup_csv="C:\path\to\setup.csv"

• --image_dims: Specifies the dimensions for the generated images in the format width,height.

  Example: --image_dims=200,200

Optional Arguments

• --no_cluster: Skip the clustering step if specified.

  Example: --no_cluster

• --no_heatmap: Skip the heatmap generation step if specified.

  Example: --no_heatmap

• --n_clusters: Specifies the number of clusters for KMeans (default is 3).

  Example: --n_clusters=5

• --no_transform: Disables square root transformation of data.

  Example: --no_transform

• --scaling_method: Choose the scaling method for clustering. Options are robust and minmax. Default is robust.

  Example: --scaling_method=minmax

• --correlation_method: Select the desired correlation method for the heatmap. Options are pearson and spearman. Default is spearman.

  Example: --correlation_method=pearson

Example Usage

Linux/MacOS:

python pixel_script.py --data_csv="/path/to/data.csv" \
                       --setup_csv="/path/to/setup.csv" \
                       --image_dims="100,100" \
                       --no_cluster \
                       --no_heatmap \
                       --n_clusters=4

Windows:

python pixel_script.py --data_csv="C:\path\to\data.csv" ^
                       --setup_csv="C:\path\to\setup.csv" ^
                       --image_dims="100,100" ^
                       --no_cluster ^
                       --no_heatmap ^
                       --n_clusters=4

Replace placeholders like path/to/csv/data.csv or C:\path\to\data.csv with your actual paths and adjust the values as needed.

Output

The script generates output in a subfolder named output within the directory containing the data_csv file. The output includes:

• clustered_image.png: An image where each pixel is colored according to its cluster.
• correlation_heatmap.png: A heatmap indicating the correlation between different channels.
• statistics.txt: A text file containing statistics related to the clusters.

Troubleshooting

KeyError or IndexError

If you encounter a KeyError or IndexError, make sure that the channels specified in setup.csv exist in data_csv.