The aim of the experimental validation group of GENCODE is to identify gene models that have limited or lower confidence transcribed evidence and to systematically experimentally validate them. Predicted exon–exon junctions were evaluated by RT-PCR amplification in eight different tissues followed by highly multiplexed sequencing readout, a method referred to as RT-PCR-seq (Howald et al. 2012). Eighty-two percent of all assessed junctions (n = 5871) are confirmed by this evaluation procedure, demonstrating the high quality of the annotation reached by the GENCODE gene set. RT-PCR-seq was also efficient at screening gene models predicted using the HBM RNA-seq data. We validated 73% of these predictions, thus confirming 1168 novel genes, mostly noncoding, which will further complement the GENCODE annotation (Howald et al. 2012). Our RT-PCR-seq–targeted approach can also be exploited to identify novel exons. We discovered unannotated exons in ∼10% of assessed introns. We thus estimate that at least 18% of loci contain as yet unannotated exons.
