PCA of Genomic Data — Namib/Angola Project

Project Description

An educational project focused on analyzing genomic data using Principal Component Analysis (PCA).
The goal is to reproduce and visualize the PCA results from the publication:
Oliveira S. et al., Genome-wide variation in the Angolan Namib Desert reveals unique pre-Bantu ancestry, Sci. Adv. 9 (2023)

---

Analysis Pipeline

1. Data Preprocessing (via PLINK)

   • SNP filtering
   • Removing individuals with low call rate
   • LD pruning
   • Selecting autosomal SNPs only

2. PCA Analysis

   • Computing PCs (PC1–PC20)
   • Explained variance ratios
   • Visualization: PC1 vs PC2, PC2 vs PC3, etc.

3. Visualization & Interpretation

   • Coloring samples by population
   • Overlaying geographical coordinates (synthetic and/or from supplementary materials)
   • Checking correlations between PCs and coordinates
   • Clustering in PC space (PyCaret, KMeans, etc.)

---

Results

• PCA clusters broadly reflect geographical population structure.
• Clustering confirms separation into groups consistent with ethnic and regional divisions.
• Some plots (especially maps and interactive visualizations) may render correctly only inside Jupyter or Colab environments.

---

How to Run

1. Install dependencies:

   pip install -r requirements.txt

2. Launch the notebook locally:

   jupyter notebook Practice_PCA_genotypes_final.ipynb

---

Tools & Libraries

• Python 3.10+
• PLINK v1.9
• pandas, numpy, scikit-learn
• matplotlib, seaborn, plotly, folium
• pycaret (for clustering)

---

⚠ Note

The dataset was provided as pre-filtered (LD-pruned) as part of the course.
All major preprocessing steps are still included in the notebook for clarity and reproducibility.