Coral Transcriptome Differential Expression Analysis Suite

A comprehensive computational pipeline for simulating and analyzing coral reef transcriptomic responses to environmental stressors, implementing standardized RNA-seq differential expression workflows with interactive visualization capabilities.

Overview

A bioinformatics tool designed for marine genomics researchers studying cnidarian transcriptional responses to environmental perturbations. The application provides a complete analytical framework for coral gene expression profiling, incorporating established methodologies from the field of marine molecular ecology and transcriptomics.

Key Features

• Transcriptome Simulation Engine: Generates realistic coral gene expression count matrices with configurable parameters mimicking RNA-seq data characteristics
• Differential Expression Analysis: Implements Welch's t-test with Benjamini-Hochberg FDR correction for identifying stress-responsive genes
• CPM Normalization Pipeline: Counts per million (CPM) transformation with log₂ scaling for variance stabilization
• Principal Component Analysis: Dimensionality reduction for sample clustering and batch effect detection
• Interactive Volcano Plots: Statistical significance visualization with log₂ fold-change thresholds
• Expression Heatmaps: Hierarchical clustering of top differentially expressed genes (DEGs)

Methodology

Data Simulation Framework

The simulation engine generates synthetic coral transcriptome data using Poisson-distributed count models:

• Baseline Expression: Control samples drawn from Poisson(λ=50) distribution
• Stress Response Modeling:
  • Upregulated genes (n=10): Additional Poisson(λ=40) counts
  • Downregulated genes (n=10): Reduced by Poisson(λ=20) counts
  • Background genes: Maintain baseline expression levels

Statistical Analysis Pipeline

1. Count Matrix Loading: Import expression data in standard CSV format with genes as rows and samples as columns

2. CPM Normalization:

   CPM = (raw_counts / library_size) × 10⁶
   log₂CPM = log₂(CPM + 1)
   

3. Differential Expression Testing:

   • Statistical method: Welch's two-sample t-test (unequal variances)
   • Multiple testing correction: Benjamini-Hochberg FDR (α = 0.05)
   • Effect size: log₂ fold-change calculation with pseudocount adjustment

4. Dimensionality Reduction:

   • Principal Component Analysis on log₂CPM-transformed data
   • Variance explained reporting for PC1 and PC2

Visualization Components

• Volcano Plot: -log₁₀(p-value) vs log₂(fold-change) with significance thresholding
• PCA Biplot: Sample ordination colored by treatment group with variance contribution
• Expression Heatmap: Z-score normalized expression of top 25 DEGs with hierarchical clustering

Installation & Dependencies

System Requirements

• Python ≥ 3.8
• 4GB RAM minimum (recommended: 8GB for large datasets)

Package Dependencies

pip install -r requirements.txt

Core Libraries:

• streamlit: Web application framework
• pandas: Data manipulation and analysis
• numpy: Numerical computing
• scipy: Statistical functions
• statsmodels: Advanced statistical modeling
• scikit-learn: Machine learning and PCA implementation
• plotly: Interactive plotting
• seaborn: Statistical data visualization
• matplotlib: Publication-quality figures

Usage

Web Application Launch

streamlit run app.py

The application will launch in your default web browser at http://localhost:8501.

Data Input Options

1. Simulated Data Generation (Default):

   • Toggle "Use existing data" checkbox OFF
   • Generates fresh mock expression matrix (200 genes × 6 samples)
   • Sample naming convention: Ctrl_1, Ctrl_2, Ctrl_3, Stress_1, Stress_2, Stress_3

2. Custom Data Upload:

   • Toggle "Use existing data" checkbox ON
   • Requires CSV format with gene identifiers as row names
   • Column headers must follow Ctrl_* and Stress_* naming pattern

Data Format Specifications

Input CSV Structure:

gene_id,Ctrl_1,Ctrl_2,Ctrl_3,Stress_1,Stress_2,Stress_3
Gene_001,45,52,48,89,76,84
Gene_002,67,61,59,45,52,48
...

Requirements:

• Gene identifiers in first column
• Sample names containing "Ctrl" for control samples
• Sample names containing "Stress" for treatment samples
• Raw count data (non-negative integers)

Interpretation Guidelines

Volcano Plot Analysis

• X-axis: log₂ fold-change (positive = upregulated in stress, negative = downregulated)
• Y-axis: -log₁₀(p-value) (higher = more statistically significant)
• Color coding:
  • Red/highlighted points: FDR-adjusted p-value < 0.05
  • Gray points: Non-significant genes

PCA Interpretation

• Sample clustering: Groups of samples with similar expression profiles
• Treatment separation: Distance between control and stress sample clusters
• Variance explained: Percentage of total expression variance captured by each PC
• Quality assessment: Tight within-group clustering indicates good biological replication

Heatmap Analysis

• Rows: Top 25 most significantly differentially expressed genes
• Columns: Individual sample replicates
• Color scale: Z-score normalized expression (red = high, blue = low)
• Clustering: Hierarchical clustering reveals gene co-expression patterns

Statistical Significance Thresholds

• p-value: Raw statistical significance from t-test
• FDR-adjusted p-value: Multiple testing corrected significance (α = 0.05)
• Fold-change: Biological effect size (typical threshold: |log₂FC| > 1)

Output Data

The application generates several data products:

1. Differential Expression Results Table:

   • Gene identifiers
   • log₂ fold-change values
   • Raw p-values
   • FDR-adjusted p-values
   • Ranked by statistical significance

2. Interactive Visualizations:

   • Volcano plot (HTML/SVG export compatible)
   • PCA biplot with variance statistics
   • Expression heatmap with clustering dendrograms

License

This project is released under the Creative Commons CC0 1.0 Universal License - see the LICENSE file for details.