IEO RNA-seq CLL Multi-Pipeline Analysis

Overview

Multi-approach transcriptomic analysis of chronic lymphocytic leukemia (CLL) response to dasatinib treatment, using data from GSE151159 (Blatte et al., 2021).

We implement three parallel pipelines on the same dataset to compare data containers, normalization strategies, and statistical frameworks for differential expression analysis:

Pipeline	Language	Data container	Normalization	DE method	Enrichment
1	R	SummarizedExperiment → DGEList	TMM + voom	limma (eBayes)	GOstats / fgsea
2	R	Seurat object	LogNormalize / SCTransform	DESeq2 (Wald)	clusterProfiler
3	Python	AnnData (scanpy)	scanpy normalize	pyDESeq2 + diffxpy	GSEApy

Repository structure

CLL_RNAseq_MultiplePipeline/
├── data/                          # Raw data (git-ignored, see setup below)
│   ├── .gitignore
│   ├── GSE151159.rds              # SummarizedExperiment with counts + metadata
│   ├── counts_matrix.csv          # Exported by 00_export_shared_data.R
│   ├── sample_metadata.csv        # Exported by 00_export_shared_data.R
│   └── gene_metadata.csv          # Exported by 00_export_shared_data.R
├── scripts/
│   ├── 01_edgeR_limma.Rmd         # Pipeline 1: SE + edgeR/limma-voom
│   ├── 02_seurat_deseq2.Rmd       # Pipeline 2: Seurat + DESeq2
│   └── 03_scanpy_pydeseq2.Rmd     # Pipeline 3: AnnData + scanpy + pyDESeq2/diffxpy
├── results/
│   ├── plots/                     # Figures (PNG/PDF)
│   ├── tables/                    # DE results, gene lists (CSV)
│   ├── enrichment/                # GO/GSEA results (HTML/CSV)
│   └── reports/                   # Knitted HTML reports
├── 00_export_shared_data.R        # Run ONCE to extract CSVs from RDS
├── IEO_RNAseq_CLL.Rproj           # R project file
├── environment.yml                # Conda env for Pipeline 3
├── renv.lock                      
├── .gitignore
└── README.md

Setup instructions

1. Clone the repository

git clone https://github.com/Sam-E18/CLL_RNAseq_MultiplePipeline.git
cd IEO_RNAseq_CLL_MultiPipeline

2. Get the data

Place GSE151159.rds in the data/ directory. This file is available and can be downloaded from GEO accession GSE151159.

3. Set up the R environment (Pipelines 1 & 2)

# In R, from the project root:
install.packages("renv")
renv::init()

# Core packages needed:
BiocManager::install(c(
  "SummarizedExperiment", "edgeR", "limma", "DESeq2",
  "sva", "GOstats", "org.Hs.eg.db", "fgsea", "clusterProfiler",
  "BiocStyle", "AnnotationDbi", "GenomicFeatures"
))
install.packages(c(
  "Seurat", "tidyverse", "pheatmap", "ggrepel",
  "knitr", "kableExtra", "rmarkdown", "here"
))

renv::snapshot()

4. Set up the Python environment (Pipeline 3)

conda env create -f environment.yml
conda activate ieo_rnaseq

5. Export shared data

# In R, from the project root:
source("00_export_shared_data.R")

This creates counts_matrix.csv, sample_metadata.csv, and gene_metadata.csv in data/ for use by all three pipelines.

6. Run the pipelines

# Pipeline 1 (R):
rmarkdown::render("scripts/01_edgeR_limma.Rmd", output_dir = "results/reports/")

# Pipeline 2 (R):
rmarkdown::render("scripts/02_seurat_deseq2.Rmd", output_dir = "results/reports/")

# Pipeline 3 (Python, with conda active):
# Can be run as Jupyter notebook or as Rmd with reticulate
conda activate ieo_rnaseq
jupyter lab scripts/03_scanpy_pydeseq2.ipynb

Dataset

• GEO accession: GSE151159
• Organism: Homo sapiens (CLL primary cells)
• Design: Responders vs. Non-responders to dasatinib (in vitro)
• Samples: 28 biological replicates (16 non-responders, 12 responders)
• Sequencing: Bulk RNA-seq

References

• Blatte et al. (2021). Gene expression profiling predicts sensitivity of CLL cells to dasatinib.
• Love MI, Huber W, Anders S (2014). DESeq2. Genome Biology 15:550.
• Ritchie ME et al. (2015). limma powers DE analyses. Nucleic Acids Research 43(7):e47.
• Robinson MD, McCarthy DJ, Smyth GK (2010). edgeR. Bioinformatics 26(1):139-140.
• Stuart T et al. (2019). Comprehensive integration of single-cell data. Cell 177(7):1888-1902.
• Wolf FA, Angerer P, Theis FJ (2018). SCANPY. Genome Biology 19(1):15.

Course resources

This project was developed as part of the IEO Transcriptomics course at UPF with some personal changes:

• DESeq2 Vignette
• edgeR User Guide
• Seurat Tutorials
• Scanpy Tutorials

Authors

Samuel Escudero, Ivon Sanchez, Karim Hamed
Universitat Pompeu Fabra