Module 5: Statistical Genomics 3 – Polygenic Prediction
This module covers critical concepts and statistical methods for polygenic prediction of complex traits and diseases. We will learn about the utility and evaluation of polygenic risk scores (PRS), the factors affecting prediction accuracy, the methodology to construct PRS using individual-level data or GWAS summary statistics, data quality controls, and challenges and pitfalls in the application. Additionally, participants will gain insights into Bayesian methods, a methodology widely used in the area, and we will introduce our in-house pipeline for polygenic prediction analysis. In practical sessions, we will use R, HPC, and command line software, including PLINK, GCTA and GCTB.
Prerequisites: If you are not familiar with GWAS, we highly recommend attending Module 1 Genetic Mapping in advance. While attending Module 3 Heritability Estimation is not necessary, a basic knowledge of statistics and programming in R and Linux is required for this module.
Goal: By the end of the module, you will acquire the skills to compute the polygenic risk scores for the trait of interest from scratch and will have a general understanding of the complex statistical methods involved in polygenic prediction.
Primary instructor:
Dr Jian Zeng
Dr Jian Zeng is a statistical geneticist and NHMRC Emerging Leadership Fellow at the Institute for Molecular Bioscience (IMB) at the University of Queensland (UQ). His research focuses on the development and application of innovative statistical methods for estimating the genetic architecture of complex traits, mapping trait-associated genetic variants and genes, and predicting trait phenotypes using genomic data.