Real-time dynamic polygenic prediction for streaming data.

Area of Science:

• Genetics
• Bioinformatics
• Precision Medicine

Background:

• Polygenic risk scores (PRSs) are vital for precision medicine but rely on outdated genome-wide association study (GWAS) data.
• Current PRS methods are suboptimal for rapidly growing genetic and health datasets, limiting prediction accuracy for new patients.

Purpose of the Study:

• To introduce real-time PRS-CS (rtPRS-CS), a novel method for dynamic PRS refinement and calibration.
• To enable continuous PRS updates as new samples are collected, avoiding the need for intermediate GWAS.

Main Methods:

• Developed rtPRS-CS for online, dynamic PRS calculation.
• Conducted extensive simulations to assess rtPRS-CS performance across diverse genetic architectures and sample sizes.
• Applied rtPRS-CS to quantitative traits from Mass General Brigham Biobank and UK Biobank, and to schizophrenia cohorts across Asian regions.

Main Results:

• rtPRS-CS effectively integrates massive streaming data to improve PRS prediction accuracy over time.
• Demonstrated the capability of rtPRS-CS to enhance PRS prediction in large-scale biobanks.
• Validated the clinical utility of rtPRS-CS in dynamically predicting and stratifying disease risk across diverse ancestries in schizophrenia cohorts.

Conclusions:

• rtPRS-CS offers a significant advancement in PRS methodology by enabling real-time adaptation to new data.
• The dynamic nature of rtPRS-CS maximizes prediction accuracy for incoming patients, advancing precision medicine.
• rtPRS-CS shows broad applicability and clinical utility for disease risk prediction across diverse populations.