AbstractOver the past decade, the majority of the mammalian genome considered to be noncoding has been revealed to be able to produce proteins. Many RNA molecules, mis‐annotated as noncoding, actually are predicted to code for proteins. Some of those proteins have been identified and verified to play critical roles in multiple biological processes. The lipid droplet (LD) is a unique cellular organelle bound with a phospholipid monolayer membrane, and is closely associated with cellular lipid metabolism and metabolic disorders. However, it is still unclear how a protein targets to LDs. Here we identified a new protein on LDs, LDANP2, which is encoded by noncoding RNA, through a proteomics‐based strategy. The key sequence for its localization on LDs, Truncation 3, is predicted to form an amphipathic helix. Surprisingly, the deletion of the first amino acid in Truncation 3 resulted in mitochondrial localization. How the types of amino acids would determine the LD or mitochondrial localizations of the protein was studied. The findings introduce a useful strategy to mine for new proteins and would provide clues to the understanding of how a protein would find its right organelle, with phospholipid monolayer or bilayer membrane.