drug treatment does not alter germline genomic variation.Meta-analysis of gene-based results is simplified since less stringent harmonization between studies is required.Multiple tissues can be evaluated using a reference transcriptome dataset (such as GTEx). In general, the only limitation is the availability of gene expression data in the given tissue for model building, which need not be, from the same study as that used for phenotype investigation. In cases where transcriptome data are available, separate analyses should be performed to simplify interpretation.The approach can be applied to common or rare variants. In general, larger sample sizes for the training set will be needed to achieve good prediction models with rare variants.
