PCAdapt

PCAdapt detects genomic loci involved in local adaptation by inferring individual-based population structure with Bayesian factor models and identifying outlier loci associated with that structure.

Key Features:

• Individual-based analysis: Uses an individual-based approach to model genetic variation without requiring predefined population labels.
• Bayesian factor models: Implements hierarchical Bayesian modeling to infer population structure through latent factors that can represent clustering or isolation-by-distance.
• Outlier locus identification: Identifies loci that exhibit atypical relationships with the inferred latent factors as candidates for local adaptation.
• False discovery rate reduction: Shows up to a two-fold reduction in false discovery rate compared with BayeScan or FST-based approaches when modeling population divergence.
• Scalability: Scales to large single nucleotide polymorphism datasets, including applications to the Human Genome Diversity Project.

Scientific Applications:

• Local adaptation detection: Identifies genomic regions under selection associated with local environmental or demographic factors.
• Population genomics studies: Infers population structure and scans for adaptive loci in population genetics datasets.
• Evolutionary inference: Supports investigations into evolutionary processes by linking genetic variation to inferred structure.
• Conservation genetics: Aids identification of adaptive genetic variation relevant to species conservation and management.

Methodology:

Hierarchical Bayesian factor models infer latent factors representing population structure (e.g., clustering or isolation-by-distance), and loci are flagged as outliers based on atypical relationships with these factors; performance benchmarks report reduced false discovery rates versus BayeScan and FST-based methods on SNP datasets such as the Human Genome Diversity Project.