Interpreting principal component analyses of spatial population genetic variation

Area of Science:

• Population genetics
• Human genetic variation
• Bioinformatics

Background:

• Principal component analysis (PCA) is widely used in population genetics.
• Cavalli-Sforza et al. pioneered PCA for human genetic variation, interpreting patterns as migration signatures.
• The interpretation of PCA patterns as migration events has been controversial.

Purpose of the Study:

• To investigate the behavior of PCA when applied to spatial genetic data with continuous variation.
• To determine if observed patterns in human genetic variation maps are true migration signals or mathematical artifacts.

Main Methods:

• Applied PCA to spatial genetic data.
• Analyzed patterns generated by PCA on data exhibiting continuous spatial variation.
• Compared observed patterns to mathematical artifacts of PCA.

Main Results:

• Gradients and waves in PCA maps of human genetic variation resemble sinusoidal mathematical artifacts.
• These patterns may arise generally when PCA is applied to spatial data.
• The observed patterns do not necessarily reflect specific migration events.

Conclusions:

• The interpretation of PCA patterns as direct evidence of migration events needs re-evaluation.
• PCA can produce artifactual patterns that mimic biological signals.
• Findings aid in the correct interpretation of PCA in population genetics and suggest its utility in correcting for continuous population structure in association studies.