Estimating narrow-sense heritability using family data from admixed populations

Area of Science:

• Population Genetics
• Quantitative Genetics
• Statistical Genomics

Background:

• Estimating narrow-sense heritability in admixed populations is complex.
• Existing linear mixed-model frameworks require evaluation in diverse populations.

Purpose of the Study:

• To evaluate linear mixed-model frameworks for heritability estimation in admixed populations.
• To propose adjustments for accurate heritability estimation in Greenlandic cohorts.

Main Methods:

• Extensive simulations were used to test different models.
• Two relationship matrices (identity by descent and identity by state) were employed.
• Principal component analysis (PCA)-based adjustment was developed and applied.

Main Results:

• A model using both relationship matrices, with PCA adjustment, yielded accurate heritability estimates in simulated data.
• Single relationship matrices resulted in biased estimates.
• Phenotypes correlating with ancestry led to inflated heritability estimates.
• Heritability for height was lower in Greenlanders compared to Europeans.

Conclusions:

• Accurate narrow-sense heritability estimation in admixed populations requires a combination of genetic relationship matrices.
• Individuals with at least one first-degree relative are crucial for reliable estimates.
• PCA-based adjustment corrects for ancestry-related inflation of heritability.