This repository contains code used for single-cell and spatial transcriptomics data analysis in the following paper: Spatial analysis of IPMNs defines a paradoxical KRT17-positive, low-grade epithelial population harboring malignant features .

How to use this repo to reproduce the analysis done in the paper?

• Data folder contains the low-size input data, ST annotations, ST dcc file paths, and samples_info spreadsheet for scRNA seq. Prior to running scripts, large data files should be downloaded as described in Data_Acquisition
• Scripts folder contains R sripts for each section of the analysis. This folder should be set to the working directory.
• Utils folder contains utility functions that are called within other scripts for a cleaner code.

---

Nanostring GeoMx Spatial Transcriptomics

Spatial transcriptomics was done using the Nanostring GeoMx platform. Segments QC was done to remove the low quality segments and genes with low detection level as recommeded by Nanostring using GeoMxWorkflows R Package, then, differential gene expression was done in 01_GeoMx_QC_Clustering_and_DGE.R. GeneSet Enrichment Analysis of cell types was performed in 02_GSEA_analysis_on_IPMN_ST.R. Extracting signatures of Normal Duct, LG-IPMN, Hyb-IPMN, HG-IPMN, and tumor as well as cell identity scoring was done using 03_GeoMx_Signatures_and_SignatureScoring.R. Comparison of PanIN and IPMN spatial transcriptomic signatures was performed in 04_Comparison_PanIN_IPMN.R

Single cell RNA Sequencing

Fastq files were aligned to hg38 using CellRanger v6.1.2. Ambient RNA correction was done using SoupX as described in 05_ambient_RNA_correction.R. Fastq files were aligned to hg38 using CellRanger v6.1.2. Samples were then merged with the scRNAseq dataset of healthy and tumor pancreas samples previously published (PMID 37021392), then integrated using the recommended Seurat rPCA MNN integration using the 06_data_integration_script.R. Clusters were then labeled using previously publised markers using 07_clusters_annotation.R. Nanostring Signatures were mapped onto the extracted epithelial cells utilizing AUCell as described in 08_ScRNA_Ductal_Analysis.R script. Inferred copy number variation analysis on the epithelial population was performed as described in 09_InferCNV.R, using ductal cells from healthy donor pancreata as the reference population.

Visium Spatial Transcriptomics

IPMN Visium post-processed and annotated data, as previously published (PMID 37285225), was downloaded from GEO accession GSE233254 in Seurat Object format and analyzed as described in 10_Visium_ST_Analysis.R script. Epithelial spots were extracted by determination of epithelial fraction from Robust Cell Type Deconvolution as greater than 0.8. Signature scoring of GeoMx-derived signatures was performed by AUCell Score.