Recent evidence from proteomics and deep massively parallel sequencing studies have revealed that eukaryotic genomes contain substantial numbers of as yet uncharacterised open reading frames (ORFs). We define these uncharacterised open reading frames as novel open reading frames (nORFs). nORFs in humans are mostly under 100 codons and found in diverse regions of the genome, including in long noncoding RNAs, pseudogenes, 3'UTRs, 5'UTRs, and alternative reading frames of canonical protein coding exons. There is therefore a pressing need to evaluate the potential functional importance of these unannotated transcripts and proteins in biological pathways and human disease on a larger scale, rather than one at a time. In this study, we outline the creation of a valuable nORFs dataset with experimental evidence of translation for the community, use measures of heritability and selection that reveal signals for functional importance in nORFs and demonstrate the potential implications for functional interpretation of genetic variants in nORFs. Our results indicate that some variants that have been previously classified as being benign or of uncertain significance may have to be reinterpreted.