Differential Gene Expression Analysis of Human Pancreatic Alpha and Beta Cells

This repository contains all R scripts and resources used to perform a full pipeline for differential gene expression (DGE) analysis using RNA-seq data from the Human Islets (HPAP) project. The aim of this project is to identify transcriptional alterations between pancreatic beta and alpha cells in type 2 diabetes (T2D) donors compared to non-diabetic individuals. This analysis is part of the Bachelor Thesis (TFG) by Carla Batlle Simó (UVic-UCC, 2025).

The project integrates public transcriptomic datasets with open-source statistical tools to:

• Process and normalize bulk and pseudobulk RNA-seq data
• Identify differentially expressed genes using DESeq2
• Visualize results (PCA, UMAP, volcano plots, heatmaps)
• Extract meaningful biological insights through enrichment analysis

The workflow includes the next steps:

---

Data Acquisition and Format

Data used in this project was downloaded from the Human Islets portal. The data is publicly available but may require prior registration or data request.

Download procedure:

1. Go to Human Islets portal
2. Navigate to the Data Download section.
3. Apply the following filter:
   • Diagnosis: SA (non-diabetic) and Type 2 Diabetes
4. Select the following datasets:
   • Bulk RNA-seq
   • Pseudobulk Alpha
   • Pseudobulk Beta
5. Download the following files:
   • donor.csv: clinical metadata per donor
   • unproc_rnaseq.csv: raw count matrix (bulk RNA-seq)
   • unproc_pbrna_Beta.csv: raw counts matrix (beta cells pseudobulk)
   • unproc_pbrna_Alpha.csv: raw counts matrix (alpha cells pseudobulk)

Required files:

• donor.csv: clinical metadata per donor
• unproc_rnaseq.csv: raw count matrix (bulk RNA-seq)
• unproc_pbrna_Beta.csv: raw counts matrix (beta cells pseudobulk)
• unproc_pbrna_Alpha.csv: raw counts matrix (alpha cells pseudobulk)

Ensure all files are placed in your R working directory (e.g., ~/Documents/practiques/data). You also have the data abailable in the repository.

---

Project Scripts

counts2tpm.R

Function to convert raw gene counts to TPM using gene length annotations.

DEG.R

Runs the DEG analysis between conditions using DESeq2. Includes data anlysis, filtering, normalization, and differential expressed genes testing. You can adjust filtering parameters inside this script.

---

Analysis Pipeline

Step-by-step:

# Metadata analysis
pca_donant_analysis("donor.csv")

# Outliers filtering
full_expression_boxplots("unproc_rnaseq.csv", "TPM_merged.csv")

# Espression distribution
distrib_bulk <- distrib_dades("unproc_rnaseq.csv")
distrib_beta <- distrib_dades("unproc_pbrna_Beta.csv")
distrib_alpha <- distrib_dades("unproc_pbrna_Alpha.csv")

# DEG Analysis
bulk_results  <- run_deg_analysis("unproc_rnaseq.csv", "donor.csv")
beta_results  <- run_deg_analysis("unproc_pbrna_Beta.csv", "donor.csv")
alpha_results <- run_deg_analysis("unproc_pbrna_Alpha.csv", "donor.csv")

# Visualization
visualize_deg_results(bulk_results,  tag = "BULK")
visualize_deg_results(beta_results,  tag = "BETA")
visualize_deg_results(alpha_results, tag = "ALPHA")

# Annotate results
get_deg_results(beta_results$res_deg)
get_deg_results(alpha_results$res_deg)

---

Parameters and Settings

The following thresholds and options are used by default (modifiable):

• Adjusted p-value (padj) threshold: 0.05
• Log2 fold change (log2FC) threshold: 1 (absolute)
• Group comparison: None vs Type2
• Genes used in plots: filtered DEGs only

All thresholds and filtering criteria are adjustable directly in the R scripts:

• Age filters: donorage >= 10 or donorage >= 18
• Zero-expression filter: max 80% of zeros per condition
• Exclude samples manually: e.g., c("R032", "R036", "R233", ...)

---

Functional Enrichment Analysis

You can take the list of DEG symbols and perform enrichment analysis using external tools:

• Enrichr: for GO terms, KEGG, WikiPathways...

  The ones used in the bachelo's thesis:

• WikiPathways: for pathway enrichment

• ChEA/ENCODE (via Enrichr): for transcription factor target enrichment

Copy the output gene lists (e.g., clean_list_beta) and paste them into Enrichr.
You may use the full list, or just upregulated/downregulated genes.
Outputs can reveal transcription factors (via ChEA/ENCODE), enriched pathways, or disease links.

---

Contact and Contributions

If you find any issues, have questions, or would like to collaborate, feel free to open an issue or submit a pull request. This project is open to improvement and extension.

---

Author

Carla Batlle Simó
Bachelor's Degree in Biotechnology - UVic-UCC (2025)

This README is part of her Bachelor's Thesis and designed to ensure reproducibility and reusability for other researchers.