Overcoming the winner's curse: estimating penetrance parameters from case-control data

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Genomewide association studies (GWAS) are crucial for identifying genetic loci influencing complex traits.
• Post-association analysis of variant penetrance and allele frequency is vital for study planning.
• Ascertainment bias, the 'winner's curse,' leads to overestimation of genetic effects in initial GWAS data.

Purpose of the Study:

• To develop and validate a statistical approach to correct for ascertainment bias in GWAS.
• To provide accurate estimates of variant frequency and penetrance parameters.
• To enhance the design and success rate of replication studies.

Main Methods:

• Developed a novel statistical method to adjust for the 'winner's curse' bias.
• Utilized simulated datasets to evaluate the performance of the correction method.
• Assessed the impact of sample size and original study power on estimate accuracy.

Main Results:

• The proposed method significantly reduces bias in parameter estimates (variant frequency, penetrance).
• Effective bias correction is achievable even with low-powered initial association studies.
• Estimate uncertainty decreases with larger sample sizes, irrespective of initial association test power.

Conclusions:

• The developed method provides reliable estimates of genetic parameters by correcting for ascertainment bias.
• Accurate parameter estimation is critical for successful replication of GWAS findings.
• Applying this method to case-control data can substantially improve replication study design and efficiency.