Accurate annotation of human protein-coding small open reading frames

Functional protein-coding small open reading frames (smORFs) are emerging as an important class of genes. However, the number of translated smORFs in the human genome is unclear because proteogenomic methods are not sensitive enough, and, as we show, Ribo-seq strategies require additional measures to ensure comprehensive and accurate smORF annotation. Here, we integrate de novo transcriptome assembly and Ribo-seq into an improved workflow that overcomes obstacles with previous methods, to more confidently annotate thousands of smORFs. Evolutionary conservation analyses suggest that hundreds of smORF-encoded microproteins are likely functional. Additionally, many smORFs are regulated during fundamental biological processes, such as cell stress. Peptides derived from smORFs are also detectable on human leukocyte antigen complexes, revealing smORFs as a source of antigens. Thus, by including additional validation into our smORF annotation workflow, we accurately identify thousands of unannotated translated smORFs that will provide a rich pool of unexplored, functional human genes. An improved workflow combining de novo transcriptome assembly and Ribo-seq validated by cellular antigen display is developed to maximize small peptide discovery, leading to identification of thousands of unannotated protein-coding smORFs.