Next Cell State Analysis

A comprehensive single-cell RNA sequencing analysis pipeline for studying cell state transitions using CellRank and Palantir algorithms.

Overview

This project analyzes single-cell RNA-seq data to understand cell state transitions and developmental trajectories. The analysis focuses on immature pyramidal neurons across different developmental timepoints (E16.5, P0, P5, P23) using advanced trajectory inference methods.

Features

• Comprehensive Preprocessing Pipeline: Quality control, normalization, feature selection, and dimensionality reduction
• Trajectory Inference: Pseudotime analysis using Palantir algorithm
• Cell State Prediction: CellRank analysis for predicting cell fates and terminal states
• Visualization: UMAP embeddings and trajectory plots
• Reproducible Environment: Conda environment with all dependencies

Dataset

The analysis uses the GSE104323 dataset, specifically focusing on:

• Cell Type: Immature pyramidal neurons
• Time Points: E16.5, P0, P5, P23 (embryonic to postnatal development)
• Sample Size: ~4,500 cells after quality control

Installation

Using Conda (Recommended)

# Create environment from the provided environment.yml
conda env create -f environment.yml

# Activate the environment
conda activate cellrank

Using pip

# Install from requirements.txt
pip install -r requirements.txt

Usage

Basic Analysis

1. Load and preprocess data:

import scanpy as sc
from notebooks.GSE104323_analysis import preprocess_time_series_data

# Load your data
adata = sc.read("path/to/your/data.h5ad")

# Preprocess
adata_processed = preprocess_time_series_data(
    adata, 
    time_col="development_stage", 
    celltype_col="cell_type"
)

2. Run pseudotime analysis:

from notebooks.GSE104323_analysis import pseudotime

# Compute pseudotime
adata_with_pseudotime = pseudotime(adata_processed, early_cell='your_early_cell_id')

3. CellRank analysis:

from notebooks.GSE104323_analysis import run_cellrank_analysis

# Run CellRank
results = run_cellrank_analysis(adata_with_pseudotime)

Key Functions

• preprocess_time_series_data(): Comprehensive preprocessing pipeline
• visualize_preprocessing(): Diagnostic plots for quality control
• pseudotime(): Palantir-based pseudotime computation
• run_cellrank_analysis(): CellRank analysis with visualization

Project Structure

next_cell_state/
├── notebooks/
│   └── GSE104323_analysis.ipynb    # Main analysis notebook
├── environment.yml                 # Conda environment specification
├── requirements.txt               # Python package requirements
└── README.md                      # This file

Dependencies

Core Libraries

• scanpy: Single-cell analysis in Python
• cellrank: Cell fate analysis
• palantir: Pseudotime inference
• scvelo: RNA velocity analysis
• pandas: Data manipulation
• numpy: Numerical computing
• matplotlib: Plotting
• seaborn: Statistical visualization

Analysis Pipeline

1. Quality Control: Filter cells and genes based on standard metrics
2. Normalization: Log transformation and scaling
3. Feature Selection: Highly variable gene identification
4. Dimensionality Reduction: PCA and UMAP
5. Clustering: Leiden clustering
6. Trajectory Analysis: Palantir pseudotime computation
7. Fate Prediction: CellRank analysis for terminal states

Results

The analysis produces:

• UMAP embeddings colored by time points, cell types, and clusters
• Pseudotime trajectories showing developmental progression
• CellRank fate probabilities and terminal state predictions
• Quality control metrics and diagnostic plots

Citation

If you use this analysis pipeline, please cite the relevant papers:

• CellRank: Lange, M. et al. "CellRank for directed single-cell fate mapping." Nature Methods (2022)
• Palantir: Setty, M. et al. "Characterization of cell fate probabilities in single-cell data with Palantir." Nature Biotechnology (2019)
• Scanpy: Wolf, F.A. et al. "SCANPY: large-scale single-cell gene expression data analysis." Genome Biology (2018)

License

This project is open source. Please check the individual package licenses for dependencies.

Contact

For questions about this analysis, please open an issue in this repository.