pcadapt: an R package to perform genome scans for selection based on principal component analysis.

Area of Science:

• Population genomics
• Molecular ecology
• Bioinformatics

Background:

• Detecting genes under selection is crucial for understanding adaptation.
• Existing methods often struggle with complex population structures and admixed individuals.
• The R package pcadapt offers a fast and scalable solution for genome scans.

Purpose of the Study:

• To introduce and evaluate the latest version of the pcadapt R package.
• To compare the performance of pcadapt with other genome scan methods.
• To highlight the utility of pcadapt for analyzing next-generation sequencing data.

Main Methods:

• Genome scans using principal component analysis (PCA) to infer population structure.
• Application of robust Mahalanobis distance as a new statistic for detecting selection outliers.
• Comparison with other software (BayeScan, hapflk, OutFLANK, sNMF) using simulated data.

Main Results:

• pcadapt demonstrates a low false discovery rate (around 10%) across various simulations.
• Mahalanobis distance in pcadapt is more powerful than the communality statistic in hierarchical structures.
• pcadapt is robust to admixed individuals, unlike BayeScan, and shows high power in population divergence and range expansion scenarios.

Conclusions:

• The updated pcadapt package, particularly with Mahalanobis distance, is a powerful and robust tool for genome scans.
• It effectively handles complex population structures, including admixed individuals and next-generation sequencing data.
• pcadapt is a valuable asset for molecular ecologists studying adaptation and population genomics.