SpatioloJI 0.1.4

A spatial transcriptomics toolkit

Spatial Transcriptomics Analysis Toolkit

This toolkit provides a comprehensive set of Python modules for analyzing spatial transcriptomics data, enabling researchers to visualize and interpret gene expression in a spatial context.

Table of Contents

• Overview
• Main Components
• Spatioloji Class
• Visualization Functions
• Spatial Analysis
• Quality Control
• Usage Examples

Overview

The Spatial Transcriptomics Analysis Toolkit is designed to handle, analyze, and visualize spatially-resolved gene expression data. It supports operations across multiple fields of view (FOVs), offers various visualization techniques, enables spatial relationship analysis, and includes robust quality control methods.

Main Components

The toolkit consists of three main Python modules:

1. Spatial_Object.py: Contains the core Spatioloji class for data management and the Spatioloji_qc class for quality control.
2. Plot_Spatial_Image.py: Provides functions for visualizing spatial data with various plotting methods.
3. Spatial_function.py: Implements spatial analysis functions such as neighbor detection and interaction analysis.

Spatioloji Class

The Spatioloji class is the core data structure that manages spatial transcriptomics data:

from Spatial_Object import Spatioloji

# Create a Spatioloji object
spatioloji_obj = Spatioloji(
    polygons=polygons_df,          # Cell polygon coordinates
    cell_meta=cell_meta_df,        # Cell metadata
    adata=anndata_obj,             # Gene expression data (AnnData)
    fov_positions=fov_positions_df # Field of view positions
)

# Load from files
spatioloji_obj = Spatioloji.from_files(
    polygons_path="polygons.csv",
    cell_meta_path="cell_meta.csv",
    adata_path="expression.h5ad",
    fov_positions_path="fov_positions.csv",
    images_folder="images/",       # Optional: folder with FOV images
)

# Save/load using pickle
spatioloji_obj.to_pickle("spatioloji_data.pkl")
loaded_obj = Spatioloji.from_pickle("spatioloji_data.pkl")

Key attributes:

• polygons: DataFrame with cell polygon vertex coordinates in local and global space
• cell_meta: DataFrame with cell metadata, including cell centers and properties
• adata: AnnData object containing gene expression data
• fov_positions: DataFrame with global coordinates of FOVs
• images: Dictionary mapping FOV IDs to image arrays
• custom: Dictionary for any user-defined additional data

Visualization Functions

The Plot_Spatial_Image.py module offers multiple visualization approaches:

1. FOV Image Stitching

from Plot_Spatial_Image import stitch_fov_images

# Create a stitched image of multiple FOVs
stitched_obj = stitch_fov_images(
    spatioloji_obj,
    fov_ids=["1", "2", "3"],  # Optional: specific FOVs to include
    flip_vertical=True,       # Whether to flip images vertically
    save_path="stitched.png", # Path to save the image
    show_plot=True            # Whether to display the plot
)

2. Global Visualization by Feature Values

from Plot_Spatial_Image import plot_global_polygon_by_features

# Plot cell polygons colored by a continuous feature
plot_global_polygon_by_features(
    spatioloji_obj,
    feature="Gene_X",               # Feature to visualize (e.g., gene name)
    background_img=True,            # Show stitched image in background
    colormap="viridis",             # Matplotlib colormap
    save_dir="./figures/"           # Directory to save output
)

# Plot cell dots colored by a continuous feature
plot_global_dot_by_features(
    spatioloji_obj,
    feature="Gene_X",               
    background_img=True,
    dot_size=20,                   # Size of dots
    colormap="viridis"             
)

3. Categorical Visualization

from Plot_Spatial_Image import plot_global_polygon_by_categorical

# Plot cell polygons colored by categorical data (e.g., cell types)
plot_global_polygon_by_categorical(
    spatioloji_obj,
    feature="cell_type",           # Column with categorical values
    background_img=True,
    color_map=None,                # None for auto-assignment or provide dict
    edge_color="black",
    alpha=0.8                      # Transparency
)

# Plot cell dots colored by categorical data
plot_global_dot_by_categorical(
    spatioloji_obj,
    feature="cell_type",
    dot_size=20,
    background_img=True
)

4. FOV-level Visualizations

from Plot_Spatial_Image import plot_local_polygon_by_features, plot_local_dots_by_categorical

# Plot multiple FOVs with continuous feature coloring
plot_local_polygon_by_features(
    spatioloji_obj,
    feature="Gene_X",
    fov_ids=["1", "2", "3", "4"],   # FOVs to visualize
    background_img=True,
    grid_layout=(2, 2)              # Optional layout as (rows, columns)
)

# Plot multiple FOVs with categorical coloring
plot_local_dots_by_categorical(
    spatioloji_obj,
    feature="cell_type",
    fov_ids=["1", "2", "3", "4"],
    background_img=True
)

Spatial Analysis

The Spatial_function.py module enables analysis of spatial relationships between cells:

from Spatial_function import perform_neighbor_analysis

# Analyze cell neighborhood relationships
results = perform_neighbor_analysis(
    polygon_file=spatioloji_obj.polygons,
    cell_metadata=spatioloji_obj.cell_meta,
    cell_type_column="cell_type",    # Column defining cell types
    distance_threshold=0.0,          # 0.0 means cells must be touching
    save_dir="./analysis/",          # Directory to save results
    include_plots=True               # Generate visualization plots
)

This analysis produces several outputs:

• Cell-cell interaction pairs
• Cell type interaction counts and statistics
• Interaction enrichment scores (observed/expected ratios)
• Visualizations including heatmaps and network graphs

Quality Control

The Spatioloji_qc class in Spatial_Object.py provides quality control functionality:

from Spatial_Object import Spatioloji_qc

# Initialize QC object
qc = Spatioloji_qc(
    expr_matrix=expr_df,         # Expression matrix
    cell_metadata=metadata_df,   # Cell metadata
    output_dir="./qc_output/"    # Output directory
)

# Run complete QC pipeline
filtered_cells, filtered_genes = qc.run_qc_pipeline()

# Or run individual QC steps
qc.prepare_anndata()
qc.qc_negative_probes()
qc.qc_cell_area()
qc.qc_cell_metrics()
qc.qc_fov_metrics()
filtered_cells = qc.filter_cells()
filtered_genes = qc.filter_genes()

The QC pipeline includes:

• Negative probe analysis (background signal control)
• Cell area and morphology assessment
• Cell-level metrics (counts, mitochondrial genes, etc.)
• FOV-level metrics comparisons
• Cell and gene filtering based on quality thresholds

Usage Examples

Complete Workflow

import scanpy as sc
import pandas as pd
from Spatial_Object import Spatioloji
from Plot_Spatial_Image import stitch_fov_images, plot_global_polygon_by_categorical
from Spatial_function import perform_neighbor_analysis

# 1. Load data
spatioloji_obj = Spatioloji.from_files(
    polygons_path="polygons.csv",
    cell_meta_path="cell_meta.csv",
    adata_path="expression.h5ad",
    fov_positions_path="fov_positions.csv",
    images_folder="images/"
)

# 2. Create stitched image of all FOVs
stitched_obj = stitch_fov_images(
    spatioloji_obj,
    save_path="stitched_image.png"
)

# 3. Visualize cell types across the tissue
plot_global_polygon_by_categorical(
    stitched_obj,
    feature="cell_type",
    background_img=True,
    save_dir="./figures/"
)

# 4. Analyze cell-cell interactions
interaction_results = perform_neighbor_analysis(
    polygon_file=stitched_obj.polygons,
    cell_metadata=stitched_obj.cell_meta,
    cell_type_column="cell_type",
    save_dir="./analysis/"
)

# 5. Access and use the results
enrichment_df = interaction_results['enrichment_df']
print("Top cell-cell interactions:")
print(enrichment_df.unstack().sort_values(ascending=False).head(10))

This toolkit enables comprehensive spatial transcriptomics analysis, from data management and visualization to sophisticated spatial relationship investigations. I will continue to add more features to it.