Transcriptomic Characterization of the Tumor Microenvironment and Chromatin Remodeler BPTF in Pancreatic Ductal Adenocarcinoma

This repository contains the code developed for my bioinformatics Master's final project. It is subdivided into four directories, following the four objectives of my research. I've also developed some useful resources stored in the 'utils' directory.

TCGA CLUSTERS ANALYSIS

This section presents the results of the RNA-Seq data analysis and survival analysis for a cohort of pancreatic ductal adenocarcinoma (PDAC) patients. The analysis is organized into 5 notebooks:

• Clustering: Unsupervised Gaussian classification of patients based on their gene expression.

• Kaplan-Meier_TCGA_clusters: Survival analysis of the clusters.

• Differential_expression_and_Functional_analysis: Differential expression and functional analysis between the clusters.

• TumorDecon: Deconvolution of immune cells and comparison of immune infiltration in both clusters.

• WGCNA: Analysis of co-expressed gene networks.

sc_RNA_Seq

Re-analysis of a sc-RNA-Seq dataset obtained from Peng et al. (Cell Research, 2019, 29:725–738; DOI). The data is not available due to its large size, but the code includes links to the repositories where it was obtained. It is divided into 5 notebooks:

• Single_Cell_RNA_Seq_analysis: Preprocessing, normalization, integration, and annotation of the sc-RNA-Seq experiment.

• Differential_expression_analysis: Analysis of differential and functional expression of ductal cells vs. malignant ductal cells and endothelial cells from tumor tissue vs. control.

• TF_Pathways_activities: Analysis of biological pathway activity and transcription factor activity in ductal cells and malignant ductal cells using decoupleR.

• Complementary_analysis: Generation of plots of interest.

• Risk_signature_analysis: Analysis of genes of interest from the risk signature.

RNA_Seq

Analysis of the transcriptomic effects (RNA-Seq) of BPTF gene silencing in conjunction with TNFa treatment in PDAC model cell lines, with samples generated in our laboratory. It contains a notebook within the RNA-Seq pre-analysis folder with the bash commands used in the RNA-Seq analysis pipeline. It also contains another notebook (RNA_Seq_Analysis) with downstream analysis.

Risk_signature

Obtaining a risk signature of BPTF-dependent genes. It contains two notebooks:

• Risk_signature: Obtaining and validating the risk signature.

• Risk_signature_genes_analysis: Analysis of the prognostic value of each gene in the signature individually.

Utils:

This folder contains four Python scripts and a pickle file, usefull resources also developed during the project:

• Volcanoplot: Generates a customized volcano plot (including highlighted gene names) from the results of differential expression.

• Functional_analysis: Contains useful functions for functional analyses, result filtering, and plot generation.

• gmt_tools: Contains functions for reading and generating .gmt files, a format commonly used for storing molecular signatures.

• gtf_dict_GRCh38.110: Contains a dictionary with the correspondence of ENSEMBL_ID and gene symbol.

• Risk signature: Contains functions necessary to obtain the risk signature.

Summary of the project findings:

Pancreatic ductal adenocarcinoma (PDAC) is a disease of increasing incidence and low survival rate (11%), due to the difficulties in its diagnosis and the lack of effective treatments. The tumour microenvironment (TME) is key in the development of this disease and in its resistance to chemotherapy and immunotherapy. In this project we have used public databases with transcriptomic and clinical information of PDAC patients to carry out a description of the TME. A multi-omics strategy with a dual approach has been applied. From a population point of view, RNA-Seq data from cohorts of PDAC patients have been analysed using machine learning tools, classical bioinformatics pipelines and network analysis. Moreover, single cell resolution was achieved by analysis of sc-RNA-Seq data. Thanks to this characterization, infiltration and activation of immune system cells were associated with increased survival. It also allowed the identification of cell types and expression patterns that lead to a state of immunosuppression in TME. Additionally, the potential of the BPTF gene as a therapeutic target was studied, proving that its silencing causes a proliferative arrest mediated by the BPTF/MYC and KRAS/PI3K/AKT/MTORC1 axes, which regulate the reprogramming of tumour metabolism. It is also proposed that BPTF silencing can reduce the inflammatory response in tumour cells by modulating the expression of TNFα-induced genes. Furthermore, BPTF-dependent genes were used to generate a risk signature with prognostic value; trained and validated with expression data from four independent patient cohorts. This approach also revealed the prognostic value of lysozyme and its possible role as a marker and therapeutic target in PDAC, which opens the door to a new line of research.

Keywords: PDAC, pancreas, tumour microenvironment, sc-RNA-Seq, BPTF, RNA-Seq, risk signature.