LDAK accepts files in three formats: binary PLINK (bed), gen and Sparse Partitioning (SP). File Formats describes the nature of these formats, and how to convert between these. Data Filtering explains how to perform QC and data manipulation on the fly.
You can use LDAK to calculate genetic risk profile scores, the projection of a dataset onto axes of effect sizes. LDAK can handle both hard genotype (0/1/2) and dosage (imputed) data, and can also calculate multiple risk scores at once.
LDAK can generate phenotypes with defined heritability and numbers of causal variants, using (if desired) pre-specified causal predictors and effect sizes. It can also generate basic (uncorrelated) SNP data. See Simulations.
At the moment, LDAK can perform basic set-based REML analysis, as used in Adaptive MultiBLUP, which enables gene-based and regional tests of association.
Advanced Options provides some (technical) options, such as defining the working directory and changing some parameter settings when calculating weightings. Tips provides some responses to questions we've been asked.