DNA-RNA_Project

Table of contents

• Overview
• Requirements
• Workflow
  • 1. Data Preparation and Import
  • 2. Signal Extraction
  • 3. Quality Control
  • 4. Beta and M Value Calculation
  • 5. Normalization and Batch Effect Analysis
  • 6. Statistical Analysis
  • 7. Visualization
• Contacts

Overview

This repository contains the final project of Group 5 for the DNA/RNA Dynamics course (MSc in Bioinformatics, University of Bologna, a.y. 2024/2025). It features a complete pipeline for analyzing Illumina 450K methylation data using R. The pipeline includes preprocessing (with the preprocessNoob method), quality control, normalization, statistical analysis (using Mann-Whitney U test and P-value threshold of 0.05), principal component analysis (PCA), and the identification of differentially methylated positions (DMPs) between control (CTRL) and disease (DIS) samples.

Requirements

The project was performed using R and Rstudio, analysing data from the platform Illumina HumanMethylation450k (input_data). Necessary packages were installed in our R enviroment:

install.packages(minfi)
install.packages(ggplot2)
install.packages(nitr)
install.packages(BiocManager)
install.packages(factoextra)
install.packages(cluster)
install.packages(qqman)
install.packages(gplots)

Project Workflow

This document outlines the step-by-step workflow of our DNA/RNA methylation project, comparing CTRL and DIS sample groups using Illumina microarray data.

---

1. Data Preparation and Import

• Install and load required R packages: minfi, BiocManager, knitr.
• Clean the R environment and set the working directory.
• Load the sample sheet using read.metharray.sheet() to import metadata.
• Read raw data using read.metharray.exp() to generate the RGset object.

---

2. Signal Extraction

• Extract fluorescence intensity data for Red (Cy5) and Green (Cy3) channels from RGset using getRed() and getGreen().
• Store the data into two separate dataframes: Red and Green.

---

3. Quality Control

• Classify sample quality based on the percentage of failed probes:
  • High quality: < 0.01%
  • Good quality: < 0.2%
  • Low quality: > 0.2%
  • Critical quality: around 1% (may require exclusion)

---

4. Beta and M Value Calculation

• Split samples into CTRL and DIS groups using metadata.
• Create MSet.raw objects for each group.
• Compute Beta values with getBeta() and M values with getM().
• Calculate and plot mean methylation values for both groups.
• Normalise data and asses the quality of normalisation.

---

5. Normalization and Batch Effect Analysis

• Perform Principal Component Analysis (PCA) to assess batch effects (e.g., by Sentrix_ID).
• PCA suggested that normalization did not fully correct batch effects.

---

6. Statistical Analysis

• Perform Mann whitney U test for each probe comparing CTRL vs DIS.
• Create a dataframe containing p-values.
• Filter probes with p ≤ 0.05 for significance.
• Plot the distribution of p-values.

---

7. Visualization

• Generate a heatmap using the top 100 most significant probes.

• Apply hierarchical clustering.

  ---

Contacts

Project members:

-Marco Cuscunà ([email protected])

-Marco Centenaro ([email protected])

-Michele Carbonieri ([email protected])

-Marina Mariano ([email protected])

-Luca Cagnini ([email protected])

-Massimo Lanari ([email protected])