Revisit Population-based and Family-based Genotype Imputation

Area of Science:

• Genetics
• Genomic Imputation
• Complex Disease Research

Background:

• Genome-Wide Association (GWA) studies using population-based imputation (PBI) have identified common variants for complex diseases, but a significant portion of heritability remains unexplained.
• Low-frequency variants (LFV) may contribute to this missing heritability, and family studies are more efficient than studies of unrelated individuals for their identification due to increased frequency within families.
• Family-based imputation (FBI) offers a promising approach to evaluate LFV.

Purpose of the Study:

• To compare the imputation performance of PBI and FBI for identifying low-frequency variants (LFV) in complex disease studies.
• To evaluate different imputation methods, including PBI, FBI, and hybrid approaches, using extensive simulations.

Main Methods:

• Extensive simulations were conducted using SeqSIMLA to generate genotypes from various reference panels for families.
• Genotype data for variants not present in Framingham 550K GWA data were masked in selected subjects.
• Imputation methods evaluated included Impute2_duoHMM for PBI, MERLIN and GIGI for FBI, and PedBLIMP for a hybrid approach.

Main Results:

• Family-based imputation (FBI) using MERLIN and GIGI generally outperformed other methods, achieving imputation accuracy >0.99 for squared correlation and imputation quality scores (IQS), particularly for low-frequency variants (LFV).
• MERLIN's imputation accuracy was influenced by pedigree splitting in larger families.
• Population-based imputation (PBI) performed less effectively, except for common variants when using a closely ancestry-matched reference panel.

Conclusions:

• Linkage disequilibrium (LD) information from large genotype resources enables good imputation of common variants with appropriate reference panels, without requiring dense sequencing in family members.
• Family-based imputation (FBI) is more effective for imputing low-frequency variants (LFV) by utilizing inheritance patterns within families, leading to improved imputation accuracy.
• FBI represents a valuable strategy for uncovering the genetic architecture of complex diseases by identifying LFV.