Definition of a Protein Complex
A cellular component should include more than one gene product; complexes of one gene product with a cofactor, e.g. heme and chlorophyll, should not
be included. Homomultimeric proteins, e.g. the homodimeric alcohol dehydrogenase, may be included as cellular component terms, as should heteromultimeric proteins, e.g. hemoglobin with alpha and beta chains.
All complexes in the component ontology should be given parentage under the general term protein complex ; GO:0043234
To distinguish cellular components from functions, use 'complex' in the term name of a component, and append enzyme names with the word 'activity'. For example, the molecular function term pyruvate dehydrogenase activity ; GO:0004738
describes the enzyme activity whereas the cellular component term pyruvate dehydrogenase complex ; GO:0045254
describes the multi-subunit structure in which the enzyme activity resides.
As a rule, GO terms to indicate association of a receptor with its ligand should not be created, as their complex may not always be stable, and there could be a potential explosion of terms. However, we should allow for exceptions. The IntAct database wouldn't curate receptor-ligand complexes if these consisted of a single chain of each, but it will curate complexes when the ligand is not monomeric and receptors oligomerize upon ligand binding. An example of this is GO:1990270 'platelet-derived growth factor receptor-ligand complex'
, where the ligands are always dimeric and the receptor dimerizes upon ligand binding. (Also, in the case of GO:1990270, the complex has been shown to exist in a variety of experiments, including crystals, pull downs, comigrations, competition assays and more.)