The Reference Sequence (RefSeq) database[1] is an open access, annotated and curated collection of publicly available nucleotide sequences (DNA, RNA) and their protein products.

[2][3] This database is built by National Center for Biotechnology Information (NCBI), and, unlike GenBank, provides only a single record for each natural biological molecule (i.e. DNA, RNA or protein) for major organisms ranging from viruses to bacteria to eukaryotes.

For each model organism, RefSeq aims to provide separate and linked records for the genomic DNA, the gene transcripts, and the proteins arising from those transcripts.

[5] RefSeq collection comprises different data types, with different origins, so it is necessary to establish standard categories and identifiers to store each data type.

Several projects to improve RefSeq services are currently in development by the NCBI, often in collaboration with research centers such as EMBL-EBI: According to the RefSeq release 213 (July 2022), the number of species represented in the database by counting distinct taxonomic IDs are as follows:[4] The counts of accession and basepairs per molecule type are:[4]