LDAK contains tools for constructing linear prediction models (polygenic risk scores) using either individual-level data or GWAS summary statistics. The two key advantages of our tools are computational efficiency and that they allow the user to specify the Heritability Model.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
When analysing individual-level data, we recommend using Elastic-Predict (however, it is still possible to use the 2021 tools Ridge-Predict, Bolt-Predict and BayesR-Predict, or even the 2014 tool MultiBLUP).
When analysing summary statistics, we generally recommend using GigaPRS, however, some users may prefer MegaPRS (GigaPRS is easier to run, whereas MegaPRS is a bit more flexible). Both tools require that your summary statistics are in the format required by LDAK (see Summary Statistics for details). Meanwhile MegaPRS also requires that you have an ancestrally-matched Reference Panel.
In Worked Examples we provide three real-data examples of computing polygenic scores using GigaPRS (the first two use single-ancestry data, while the third uses multi-ancestry data), as well as an example of using MegaPRS.
