Team2_AI_Microbiome_Poultry

A collaborative BioAI Hackathon project applying machine learning to poultry microbiome data for predictive health and production trait modeling.

---

📌 Team

• Mxolisi Nene (Team Lead)
• Karthikeyan Govindan
• Rajarshi Mondal

---

🎯 Goal

Use publicly available poultry microbiome datasets (16S rRNA / Metagenomics) to identify microbial biomarkers associated with growth, gut health, and disease resistance using AI techniques.

---

🛠️ Methodology

• Data preprocessing & normalization
• Microbiome taxonomic profiling using:
  • Shotgun metagenomics: Kraken2
  • 16S rRNA amplicon sequencing: DADA2
• ML techniques:
  • Clustering
  • Classification
  • Feature selection
• Results interpretation & visualization

---

📦 Deliverables

• AI-informed microbiome health prediction model
• Presentation & report for BioAI Hackathon 2025

---

📄 License

MIT

---

📊 Workflow Diagram

flowchart TD
  A[Download SRA] --> B[QC & Trim]
  B --> C{Shotgun?}
  C -->|Yes| D[Kraken2 Profiling]
  C -->|No| E[DADA2 16S Processing]
  D --> F[Generate Taxa Table]
  E --> G[Generate Feature Table]
  F --> H[Model Training]
  G --> H
  H --> I[Report]

Loading

---

📥 Data Retrieval and Preprocessing

📜 Script: data_retrieval.sh

This script automates the process of:

1. Downloading sequencing data from ENA using a BioProject ID
2. Performing adapter trimming using fastp
3. Generating a MultiQC report for quality control

---

▶️ Usage

To run the script, use the following command:

./data_retrieval.sh [BioProject_ID]

🔍 Example

./data_retrieval.sh PRJNA707106

⚙️ Data Processing Pipeline Using DADA2 in R

📜 Script: dada2_script.R

This script automates the essential preprocessing steps for amplicon sequencing data, including:

✂️ Quality filtering and primer trimming to clean raw reads

🔗 Merging paired-end reads for accurate sequence reconstruction

🚫 Removing chimeras to ensure data integrity

🎯 Generating the final Amplicon Sequence Variant (ASV) table ready for taxonomic classification

🧬 Taxonomic Annotation

Taxonomy is assigned using the SILVA reference database:
silva_nr99_v138.1_wSpecies_train_set.fa.gz

Outputs include taxonomic tables at the Genus and Species levels

Results are saved as CSV files, paired with metadata for easy integration

🚀 Ready for Downstream Analysis

The processed data is perfectly formatted to jump straight into:

Machine learning workflows

Visualization and reporting