OSCC RNA-seq analysis

This repository contains a differential expression analysis of oral squamous cell carcinoma (OSCC) using DESeq2, followed by pathway enrichment analysis using GSEA and Pathweigh. The project investigates transcriptomic changes across three sample classes: Normal, Dysplasia, and OSCC.

---

Contents

• deseq2_all_paired.Rmd: R Markdown file with DESeq2 analysis for paired samples.
• results.pptx: Summary presentation of pathway enrichment results and class comparisons.

---

Analysis Overview

The complete transcriptomic analysis pipeline includes the following steps:

1. Combine Lanes
   Merge raw sequencing reads from different lanes for each sample.

2. FastQC
   Perform quality control checks on raw reads.

3. MultiQC
   Aggregate FastQC reports into a single summary report.

4. Fastp
   Trim adapter sequences and remove PCR duplicates.

5. FastQC (post-trimming)
   Re-assess read quality after trimming.

6. MultiQC (post-trimming)
   Summarize updated FastQC results.

7. Kallisto
   Quantify transcript abundances using pseudoalignment.

8. Differential Expression with DESeq2

   • Pairwise comparisons across conditions:
     • Dysplasia vs Normal
     • OSCC vs Dysplasia
     • OSCC vs Normal
   • DESeq2 was run with a paired design (~ subject + condition).
   • Log2 fold change shrinkage applied (apeglm).

9. Pathway Enrichment

   • GSEA (MSigDB Hallmark & C4 gene sets):

     • Hallmark: 50 gene sets – Dysplasia shows transient upregulation of specific pathways.
     • C4: 858 gene sets – Persistent upregulation across Dysplasia and OSCC vs Normal.

   • Pathweigh Analysis:

     • Input: TPM-normalized transcript abundance.
     • Output: 581 pathways x sample proportions.
     • Used Kolmogorov-Smirnov tests to detect distribution shifts across classes.

---

Key Findings

• Dynamic pathway regulation: Some pathways show increase in Dysplasia and decrease in OSCC.
• Persistent upregulation: Other pathways are continuously elevated from Normal to OSCC.
• Class separation: Significant pathways (via KS test) help distinguish all three sample types visually.

---

Requirements

• R 4.2+
• DESeq2
• tidyverse
• EnhancedVolcano
• pheatmap
• RColorBrewer

To install the required packages:

install.packages("tidyverse")
if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install(c("DESeq2", "EnhancedVolcano", "pheatmap", "RColorBrewer"))