Lineage tracing of Shh+ floor plate cells and dynamics of dorsal-ventral gene expression in the regenerating axolotl spinal cord

This project involves the analysis, modeling and visualization of gene expression data from multiple genes, conditions and replicas using a piecewise constant fitting approach.

Prerequisites

The following Python packages are required:

• pandas
• numpy
• scipy
• matplotlib
• seaborn

You can install these packages using pip:

pip install pandas numpy scipy matplotlib seaborn

Data preprocessing

Loading data

The data is loaded from an Excel file (quantifications_all-n.xlsx). Each sheet in the Excel file contains distance and expression data for various genes, conditions and replicas.

Normalizing and binning data

Distances are normalized to a range of 0 to 100 and binned into specified intervals. The average expression within each bin is calculated.

Categorizing data by gene and condition

Data is organized into dictionaries based on genes and conditions, and statistical metrics are computed.

Visualizing data

• Expression profiles: plots of gene expression profiles along the distance axis for each gene and condition.
• Piecewise constant fitting: fitting the data using two-step and three-step piecewise constant functions to identify changes in expression levels.

Piecewise constant fitting

Two-step piecewise constant fitting

The two-step piecewise constant function fits the data with one switchpoint for most of the genes. The best fit is determined by evaluating the SSE for all potential switchpoints.

Three-step piecewise constant fitting

The three-step piecewise constant function is used to fit the data with two switchpoints for Pax6. The function evaluates different pairs of switchpoints to minimize the sum of squared errors (SSE).

Visualizing fitting results

• SSE plots: scatter plots showing the SSE values for different switchpoints.
• Expression plots: scatter plots of the data with the fitted piecewise constant function overlaid.

Statistical comparison

• Intensity and size comparison: box plots comparing the corrected HCR signal intensity and domain size between conditions for each gene.

Statistical annotations

Statistical tests (Mann-Whitney and t-test) are performed to compare the differences between conditions, and the results are annotated on the plots.

Summary statistics

Summary statistics for gene expression data, including mean, median and std of intensity and size, are calculated and saved to a CSV file (summary_stats.csv).

Running the analysis

To run the analysis, execute the script in a Python notebooks with the necessary packages installed. The script will generate visualizations and save the results to the specified output files.

Contact

For questions or further information, please contact [email protected].