Estimating disorder probability based on polygenic prediction using the BPC approach.

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Polygenic Scores (PGSs) estimate genetic predisposition for traits.
• Existing Bayesian PGS methods lack calibration for binary disorder traits in ascertained samples, hindering clinical use.
• Accurate calibration is crucial for implementing PGSs in healthcare settings.

Purpose of the Study:

• To introduce the Bayesian polygenic score Probability Conversion (BPC) approach.
• To provide a practical method for computing well-calibrated disorder probabilities using PGSs.
• To improve the accuracy of genetic risk prediction for binary traits.

Main Methods:

• The BPC approach utilizes genome-wide association study summary statistics, existing Bayesian PGS methods (e.g., PRScs, SBayesR), and individual genotype data.
• It incorporates a flexible prior disorder probability, adaptable from various sources.
• The method does not require a separate tuning sample with both genotype and phenotype data.

Main Results:

• BPC yields well-calibrated disorder probability predictions.
• The approach demonstrates consistently superior performance compared to another recent method.
• Validation was performed using simulated data and empirical data across nine disorder traits.

Conclusions:

• The BPC approach offers a practical and effective solution for calibrating PGSs for binary disorder traits.
• It enhances the reliability of genetic risk prediction, paving the way for clinical implementation.
• BPC improves upon existing methods, providing more accurate and trustworthy genetic insights.