In the UniProt RDF model, the accession is only in the IRI of the form http://purl.uniprot.org/uniprot/${ACCESSION}.

To go from an accession string in pathway commons to a IRI one uses a SPARQL snippet like:

VALUES ?acc { "P05067" }
BIND(IRI(CONCAT("http://purl.uniprot.org/uniprot", ?acc)) AS ?entry)

There are two reasons that we don't have the primary accession as a string in our RDF or SPARQL endpoint.

1. Avoiding false joins, an UniProt accession. Might also be used to identify something completely else, without the IRI part false joins can lead to wrong results.
2. Adding a string for each identifier adds hundreds of millions of extra triples and strings in the database which will negatively impact performance and storage.