Determining population structure from k-mer frequencies.

Area of Science:

• Genomics
• Population Genetics
• Bioinformatics

Background:

• Understanding population structure is crucial for evolutionary biology and large-scale genetic studies.
• Current methods for population structure inference often rely on genetic markers and can involve complex assumptions.
• Existing approaches include model-based, statistical, and distance-based ancestry inference.

Purpose of the Study:

• To develop and evaluate an alignment-free method for determining population structure using DNA sequence data.
• To utilize k-mer frequencies and principal component analysis (PCA) as a novel approach for population structure inference.
• To compare the performance of the k-mer based method against SNP-based methods.

Main Methods:

• Employed an alignment-free strategy utilizing the frequencies of short DNA substrings (k-mers) across the genome.
• Applied principal component analysis (PCA) to k-mer frequency profiles for population structure analysis.
• Validated the approach using both simulated data and empirical human genome data from the 1000 Genomes Project.

Main Results:

• PCA applied to k-mer frequencies successfully identified population structure, comparable to SNP-based methods in simulations.
• The k-mer based PCA approach outperformed SNP-based estimates in human genome data from the 1000 Genomes Project.
• The method demonstrated effectiveness in separating admixed and non-admixed populations.

Conclusions:

• Principal component analysis (PCA) combined with k-mer frequencies provides an effective and accessible method for population structure detection.
• This alignment-free approach circumvents genetic assumptions and marker selection challenges inherent in traditional methods.
• The k-mer frequency method shows potential for improved population structure estimates, particularly in smaller sample sizes.