Generation of ENSEMBL-based proteogenomics databases boosts the identification of non-canonical peptides