🧬 Metagenomic Analysis: Type 2 Diabetes vs. Healthy Controls

📖 Project Overview

This project investigates the differences in gut microbiomes between patients with Type 2 Diabetes (T2D) and Healthy Controls. Using 120 metagenomic samples, we built a bioinformatics pipeline to answer two fundamental questions:

1. Taxonomy: Who is there? (Bacterial composition)
2. Function: What are they doing? (Metabolic potential)

Our goal was to go beyond simple bacterial lists and explore microbial load (biomass) and functional gene abundance related to gut health.

---

🧪 Study Design

• Dataset: 120 Samples (PRJNA422434)
  • 🔴 T2D Group: 60 Samples
  • 🟢 Control Group: 60 Samples
• Method: Shotgun Metagenomic Sequencing.
• Pipeline Manager: Snakemake

---

⚙️ The Pipeline (Methodology)

We used a "Targeted Mining" approach, focusing on specific metabolic markers rather than computationally expensive whole-genome annotation.

1. Quality Control: fastp (Cleaning raw reads).
2. Profiling: Kraken2 (Identifying bacterial species).
3. Assembly: MEGAHIT (Reconstructing genomes).
4. Gene Mining: BLAST (Searching for Butyrate Kinase and BCAA genes).

---

📊 Key Results & Visualizations

1. The Taxonomic Landscape

We identified the most dominant bacteria across the cohort. The results show a microbiome heavily dominated by Segatella copri (formerly Prevotella copri) and Bacteroides species. This specific composition is often linked to "Enterotype 2" and has been associated with insulin resistance in recent literature.

Figure 1: Top 10 most abundant species across 120 samples. Note the clear dominance of Segatella copri.

---

2. The "Biomass" Discovery (Assembly Analysis)

A striking finding emerged when comparing the assembled genomes. T2D samples yielded significantly larger assemblies compared to healthy controls.

• T2D Mean Length: ~149 Mb
• Control Mean Length: ~101 Mb

This suggests a state of bacterial overgrowth (higher total microbial load) in the disease group.

Figure 2: Boxplot comparing Total Assembly Length. T2D samples (right) show a significantly higher total length compared to Controls (left).

---

3. Functional Gene Mining (The "buk" Gene)

We targeted the Butyrate Kinase (buk) gene, a marker for beneficial butyrate production.

• Result: The gene was found in 100% of samples.
• Significance: T2D patients showed a significantly higher absolute count of this gene ($p < 0.0001$).
• Interpretation: This mirrors the assembly results. T2D patients have more bacteria overall, so they have more copies of this gene. It indicates an expansion of the total community rather than a specific enrichment of beneficial bacteria relative to the rest.

Figure 3: Abundance of metabolic marker genes. The 'buk' gene (left) is significantly higher in T2D, correlating with the increased biomass.

---

💡 Conclusion

This study highlights that Type 2 Diabetes in this cohort is characterized by:

1. A specific taxonomic signature dominated by Segatella copri.
2. A significant increase in total microbial biomass (bacterial overgrowth).
3. An increase in functional gene counts driven by this biomass expansion.

These findings suggest that considering total microbial load is crucial when interpreting metagenomic data in metabolic diseases.

---

🛠️ Tech Stack

• Languages: Python, Bash
• Workflow: Snakemake
• Bioinformatics Tools: fastp, Kraken2, MEGAHIT, BLAST+
• Analysis: Pandas, Seaborn, Biopython