Netris LapNet

Transcriptomic Analysis of PDAC Tissues (LapNet Study)

---

This repository contains the R code and analysis pipeline for the manuscript:

Roth et al. (2026). Netrin-1 blockade alleviates resistance to first line chemotherapy in locally advanced pancreatic cancer. Nature.

The analysis covers the preprocessing of microBulk RNA-seq from FFPE samples, molecular subtyping, differential expression analysis (Pre vs. Post therapy), and integration with public datasets (GSE253260, GSE131050).

---

1. Project Overview

• Preprocessing: Alignment with Rsubread and quantification for FFPE-specific considerations.

• Classification: Molecular subtyping using PDACMOC, PurIST, and ESTIMATE.

• Differential Expression: edgeR pipeline for paired and unpaired analysis.

• Validation: Integration and GSEA comparison with Nicolle et al. and Linehan et al. datasets.

2. Getting Started

Prerequisites:

You will need R (>= 4.0) and the following core Bioconductor packages: edgeR, Rsubread, clusterProfiler, org.Hs.eg.db, and singscore.

Installation:

Clone this repository:

git clone https://github.com/Genomics-Consulting/Netris_LapNet.git

Open Netris_LapNet.Rproj in RStudio.

Restore the R environment using renv:

install.packages("renv")
renv::restore() 

Molecular Subtyping Setup:

1. First, create the base environment containing the R and Python dependencies:

# bash
conda env create -f env/PDACMOC_env.yml
conda activate PDACMOC

2. The PDACMOC and ADVOCATE packages must be installed from source after activating the environment. From within the directory where you have cloned the PDACMOC repository, run:

# Inside an R session
# Install the ADVOCATE dependency required for stroma classification
install.packages("inst/packages/ADVOCATE_0.1.0.tar.gz", repos = NULL, type = "source")

# Install the main PDACMOC package
install.packages(".", repos = NULL, type = "source")

3. The PDACMOC package uses Python-based machine learning models. Script 2_molecular_classification.R is configured to use the Python interpreter provided by the Conda environment:

reticulate::use_condaenv("PDACMOC")

External Data:

While data/pdata_LapNet.csv is included in this repo, the raw expression data and large objects should be downloaded from GEO:

• LapNet Data: GEO Accession GSE319924

• Public Data: GSE253260 (Nicolle et al.), GSE131050 (Linehan et al.), and GSE225691 (Cassier et al.).

3. Usage

The analysis is divided into sequential scripts. Please run them in order:

1. scripts/1_preprocessing.R: Alignment and initial QC.

2. scripts/2_molecular_classification.R: Application of PDAC-specific classifiers.

3. scripts/3_differential_Pro.vs.Pre.R: Identification of DEGs and GSEA.

4. scripts/4_public_datasets.R: Processing and integration of validation cohorts.

5. scripts/5_LapNet_pdata.R: Consolidation of metadata for final figures.

6. scripts/6_Figures.R: Generation of all main and extended figures.

4. Citation

If you use this code or the LapNet dataset, please cite:

Roth et al. (2026). Netrin-1 blockade alleviates resistance to first line chemotherapy in locally advanced pancreatic cancer. Nature.

---

📧 Contact & Collaboration

For questions or collaboration opportunities, please contact us at contact@genomicsconsulting.eu or visit our website.

Genomics Consulting SARL – 999 261 738 R.C.S. Lyon
Registered Office: 36B rue de la Batterie, 69500 Bron, France

---