New clustering methods for population comparison on paternal lineages.

Area of Science:

• Population Genetics
• Computational Biology
• Bioinformatics

Background:

• Understanding population history relies on analyzing genetic variations.
• Y-chromosomal haplogroup frequencies offer insights into paternal lineage and demographic movements.
• Existing clustering methods may have limitations in analyzing complex, high-dimensional genetic data.

Purpose of the Study:

• To introduce and evaluate two novel computational techniques for clustering population genetic data.
• To identify distinct genetic clusters within 90 Western Eurasian populations based on Y-chromosomal haplogroup frequencies.
• To assess the efficacy of the new methods in interpreting demographic history.

Main Methods:

• Development of the self-organizing cloud (SOC) algorithm, a vector-based self-learning method.
• Implementation of the maximal relation probability (MRP) algorithm, a novel probabilistic approach.
• Comparative analysis of SOC, MRP, and the k-medoids algorithm using 18-dimensional Y-chromosomal haplogroup data from 90 populations.

Main Results:

• The SOC and MRP algorithms successfully identified 10 distinct population clusters.
• These clusters exhibited strong genetic, geographic, and historical coherence.
• The identified clusters mirrored the early dispersal patterns from the Fertile Crescent across Eurasia.

Conclusions:

• The novel SOC and MRP algorithms provide an efficient and interpretable approach to population clustering.
• These methods are valuable tools for reconstructing demographic histories and understanding population relationships.
• Parallel application of SOC and MRP enhances the study of populations with shared genetic heritage.