Estimation of genotype relative risks from pedigree data by retrospective likelihoods

Area of Science:

• Genetics
• Biostatistics
• Epidemiology

Background:

• Pedigrees from linkage studies are valuable for estimating genetic relative risks (RRs).
• Genome-wide association studies (GWAS) have identified numerous genetic variants associated with diseases.
• Estimating RRs from ascertained pedigrees requires methods that adjust for ascertainment bias.

Purpose of the Study:

• To develop and evaluate methods for estimating genetic relative risks (RRs) from highly ascertained pedigrees.
• To compare the performance of a retrospective likelihood approach with a composite likelihood method.
• To apply these methods to real-world data from a prostate cancer linkage study.

Main Methods:

• Utilizing a pedigree retrospective likelihood approach to adjust for ascertainment.
• Implementing a Newton-Raphson method for computationally efficient estimation, particularly for single nucleotide polymorphisms (SNPs) with log-additive effects.
• Simulating data to compare the retrospective likelihood with a naive composite likelihood method.

Main Results:

• The retrospective likelihood method provides accurate RR estimates from ascertained pedigrees.
• A naive composite likelihood method can lead to biased RR estimates due to inadequate conditioning.
• Estimated RRs from prostate cancer pedigrees were consistent with odds ratios from case-control studies.

Conclusions:

• The retrospective likelihood method is a valid and efficient approach for estimating genetic relative risks from ascertained pedigrees.
• This method enhances the utility of existing pedigree resources for genetic research.
• The findings support the use of retrospective likelihood in genetic epidemiology.