We have developed MAGMA, a fast and flexible method for performing gene and gene-set analysis in a two-tiered parametric framework. Comparison with a number of other, frequently used methods shows that MAGMA has better power for gene analysis as well as for both self-contained and competitive gene-set analysis. An important factor in this is the multiple regression model used in the gene analysis, which is better able to incorporate the LD between SNPs than other methods. Because of its two-layer structure, this improvement in power at the gene-level subsequently carries over to the gene-set analysis.
