CellVista: Single-Cell Transcriptome Analysis with Novel Indices

Overview

CellVista is a user-friendly Shiny application designed for researchers and bioinformaticians working on single-cell RNA-seq (scRNA-seq) analysis.
The app integrates standard scRNA-seq workflows (QC, clustering, pseudotime) with three novel analysis modules:

• CellEntropy → quantifies transcriptional diversity across cells
• RTI (Regulatory Turbulence Index) → measures transcriptional instability and regulatory noise
• BSI (Boundary Sharpness Index) → a new metric that quantifies the clarity of transcriptional boundaries between clusters or states

With CellVista, users can easily explore single-cell landscapes, identify stable vs. transitional cell states, and gain deeper insights into tumor heterogeneity and microenvironment dynamics.

---

✅ Key Features

• Data input via .rds files (Seurat v5 objects)
• Automated QC & normalization
• Clustering and UMAP visualization
• Pseudotime trajectory inference
• Entropy landscape transcriptome analysis
• RTI (Regulatory Turbulence Index) for instability quantification
• BSI (Boundary Sharpness Index) for cluster boundary sharpness
• High-resolution plots and CSV export

---

📂 Input Requirements

• .rds file (Seurat v5 format)

➤ Preparing Your Data

# Example (from GEO matrix):
seurat_obj <- CreateSeuratObject(counts = your_matrix)
saveRDS(seurat_obj, file = "your_dataset.rds")

---

🧪 Datasets Tested

CellVista has been successfully tested on multiple real-world datasets:

Dataset Type	Source	Description
Lung cancer (GEO)	GEO	Primary development dataset
Breast cancer (GEO)	GEO	Validation dataset
Pancreatic cancer (GEO)	GEO	Tumor microenvironment
Brain tumor (GEO)	GEO	Glioblastoma test dataset
Hypothalamus development (GEO)	GEO	Non-cancer developmental dataset (mouse, 11 timepoints)

All datasets processed successfully, confirming robustness and generalizability.

---

🎯 Quick Start

Option 1: Run in RStudio

1. Open app.R
2. Click Run App

Option 2: R Console

shiny::runApp("your_app_folder_path")

---

📦 Demo Files Included

File Name	Description
demo_lung_seurat.rds	GEO lung cancer subsample
demo_breast_seurat.rds	Breast cancer subsample
demo_pancreas_seurat.rds	Pancreatic cancer subsample
demo_brain_seurat.rds	Glioblastoma subsample
demo_hypothalamus_seurat.rds	Mouse hypothalamus developmental subsample (non-cancer)

⚠️ These are 200-cell subsamples to meet GitHub limits. Full datasets used for testing are available on request.

---

📄 License

MIT License

---

📬 Contact

Developer: Haider
📧 haider@emails.bjut.edu.cn

---

📚 Citation

If CellVista contributes to your research, please cite this tool in your publication.