Skip to content

Integrative Analysis of Differential Gene Expression, Transcript Usage, and RNA Secondary Structure in Early- and Late-Onset Colorectal Cancer

by Salomon Marquez

11/02/2025

This site contains a summary of my bioinformatics master's thesis developed at CIMA University of Navarra under the supervision of Fernando Pastor Rodriguez and Igor Ruiz de los Mozos.

For a detailed overview of the project, you can check the following resources:

Abstract

Colorectal cancer is increasingly diagnosed in individuals younger than 50 years, a trend referred to as early-onset colorectal cancer. This study applied an integrative transcriptomic workflow to characterize molecular differences between early-onset and late-onset colorectal cancer. A total of 86 matched tumor and normal samples were analyzed using a combination of differential gene expression, differential transcript usage, and RNA secondary structure prediction.

Differential gene expression analysis identified a 40-gene signature uniquely deregulated in early-onset colorectal cancer, enriched in immune-related, inflammatory, and cell-matrix adhesion processes. Differential transcript usage analysis revealed substantial differences in transcript-level complexity between cohorts, with late-onset colorectal cancer showing extensive isoform remodeling, whereas early-onset tumors displayed a more restricted pattern of transcript usage. Despite these global differences, six conserved isoform switches were consistently detected in both cohorts, including TPM2, LMNB2, MAP3K20, and the long non-coding RNA HOXB-AS3. RNA secondary structure prediction further characterized significant isoforms, revealing a predominance of stem motifs.

These results indicate that gene-level and isoform-level analyses capture complementary aspects of colorectal cancer transcriptomic variation. The integrative framework presented here provides a methodological basis for future large-scale studies and supports the exploration of RNA structural features as potential biomarkers and therapeutic targets in colorectal cancer.

keywords: colorectal-cancer, deseq2-analysis, dtu-analysis, rna-structure-prediction, nf-core-rnaseq, isoformSwitchAnalyzeR, hpc-cluster, nextflow

Objective

The primary goal of this project is to identify genes and transcripts that are differentially regulated in early-onset and late-onset colorectal cancer, and to evaluate whether changes in isoform usage are associated with differences in their RNA secondary structure through computational prediction.

Proposed solution

We adopted an integrative transcriptomic strategy to investigate molecular differences between early-onset and late-onset colorectal cancer (EOCRC and LOCRC). Building upon a previously conducted gene-level study from Marx et al., we extend the analysis by incorporating isoform-level regulation and RNA secondary structure prediction, thereby moving beyond expression changes alone to explore post-transcriptional and structural layers of regulation. The overall integrative analysis workflow is summarized below:

tfm-workflow Schematic overview of the integrative transcriptomic analysis workflow implemented in this study.

From concept to implementation

The main milestones of this project were tracked using a dedicated 🏁 roadmap on GitHub projects, which facilitated version control and documentation of analytical decisions. Among these decisions, the main technical and methodological challenges encountered during the development of this project and the mitigation strategies adopted are listed below:

Summary of products obtained

This project produced the following deliverables.

  • A master’s thesis manuscript documenting the analytical workflow, results, and their biological interpretation.
  • A configured and validated nf-core/rnaseq Nextflow pipeline for reproducible RNA-seq analysis on high-performance computing (HPC) clusters.
  • A Zenodo repository hosting primary and processed data generated in this study.
  • A GitHub repository containing primary scripts and documentation.

Results

This is a summary of results after the implementation of the integrative transcriptomic workflow for the analysis of EOCRC and LOCRC samples.

1. nf-core/rnaseq pipeline — EOCRC

SRA study SRP357925
21 EOCRC patient pairs

2. nf-core/rnaseq pipeline — LOCRC

SRA study SRP479528
22 LOCRC patient pairs

3. Differential Gene Expression (DGE)

4. Differential Transcript Usage (DTU)

5. RNA Secondary Structure Analysis (RSS)

Contributing

This is a thesis repository, but suggestions for improvements are welcome via GitHub issues.