SeqHub Genomic Neighborhoods Now Integrated in UniProt

UniProt entries for prokaryotic proteins now show functionally similar proteins, retrieved using SeqHub's genomic language model (gLM2), in their genomic context. The context and underlying contigs come from SeqHub's OpenGenome database of 130,000+ microbial genomes and 400M+ proteins.

This means anyone looking up a protein on UniProt, whether it's well characterized or listed as a hypothetical protein, can now immediately see how functionally similar proteins are positioned in their genomes, and how well the genomic context is preserved across taxa.

UniProt is the canonical reference database for protein information, and genomic neighborhoods are a critical piece of understanding protein function, alongside sequence, structure, and existing annotations. Retrieving contextual information reliably at scale is difficult and often requires stitching together files across databases and combing through raw genome assemblies.

SeqHub makes it simple and fast to retrieve genomic context for proteins of interest. By bringing that data directly onto UniProt entries, researchers can see it without leaving the page or running their own analyses.

The panel shows the four most functionally similar proteins (from SeqHub's OpenGenome database) to your query, each in its genomic context. Every row is a genomic contig centered on one of these proteins (indicated with a pin), and ranked by degree of functional similarity (measured by embedding distance from the gLM2 model). Colored genes in each contig represent genes that commonly co-occur alongside the pinned match, and on hover, you'll see predicted annotations for each gene.

The panel is designed to enhance the UniProt entry itself, so the genomic context sits right where you're already reading. You can explore the full set of matches by clicking through to SeqHub.

For more specifics, visit UniProt's help page.