Biogeographic Ancestry Analysis of Microtia Patients in Colombia using Nonlinear Probabilistic Clustering

Area of Science:

• Population genetics
• Human ancestry
• Disease association studies

Background:

• Understanding population composition is vital for genetic association studies and disease research.
• Single nucleotide polymorphisms (SNPs) are key markers for human genetic variability.
• Accurate inference of population structure is essential for genetic analyses.

Purpose of the Study:

• To propose a hierarchical framework for nonlinear probabilistic clustering of individuals with mixed ancestry.
• To identify latent variables capturing complex genetic variation patterns.
• To infer population structure for genetic association studies.

Main Methods:

• Kernel Principal Component Analysis (Kernel PCA) for feature extraction.
• Gaussian Mixture Models (GMM) for probabilistic clustering.
• Training on diverse pure populations (Africa, Europe, East Asia, America).

Main Results:

• Achieved an adjusted Rand index of 0.981 on the evaluation set, demonstrating high accuracy.
• Validated on real datasets with a Mean Squared Error of 2.77%.
• Identified a significant association between Native American ancestry and microtia prevalence in Colombian individuals.

Conclusions:

• The proposed framework effectively clusters individuals and infers population structure.
• The method accurately identifies genetic variations relevant to population composition.
• Demonstrated the utility in uncovering population-specific disease associations, exemplified by microtia.