The Human Proteome Project currently seeks to understand the complexities of functional genomics and proteomics and to systematically identify and functionally characterize noncanonical translated sequences, such as micropeptides, will enrich the understanding of the human genome. By producing a high-coverage micropeptide sequencing reference library with 11,931,202 open reading frames and devising an alternative ultrafiltration-concentrated tandem mass spectrometry assay, we effectively confirmed 10,771 unlabeled proteins. Nearly 40% of them were derived from noncoding RNAs, suggesting that there may be a large number of misannotations in the human genome. These micropeptides exhibited abundance at subcellular locations and broad gastric cancer relevance. CRISPR screening showed that 1161 candidates had critical growth regulatory functions that were associated with their genetic properties. Excitingly, shorter sequenced micropeptides exhibited natural peptide-based drug potential or strong carcinogenic effects. Our findings illustrate a hitherto unrecognized micropeptidome and highlight its valuable implications for cancer therapy and drug development.