Expression prediction models with LD reference data are available in this website. The models were trained on Common Mind Consortium, GTEx, and STARNET consortiums. The underlying algorithm is Elastic Net, informed by epigenetic data.
These models were presented in “Integrative transcriptome imputation reveals tissue-specific and shared biological mechanisms mediating susceptibility to complex traits”, Zhaneg et al, 2019, Nature Communications. Please cite if you find these useful.