Proteins perform various functions in living organisms. Automatic protein function annotation is defined as finding appropriate association between proteins and functional labels like Gene Ontology (GO) terms. n this paper, we present a preliminary exploration of the potential of generative adversarial networks (GAN) for protein function annotation. The Prot-A-GAN approach uses GAN-like adversarial training for learning embedding of nodes and relation in an heterogeneous knowledge graph. Following the terminologies of GAN, we firstly train a discriminator using domain-adaptive negative sampling to discriminate positive and negative triples, and then, we train a generator to guide a random walk over the knowledge graph that identify paths between proteins and GO annotations. We evaluate the method by performing protein function annotation using GO terms on human disease proteins from UniProtKB/SwissProt. As a proof-of-concept, the conducted experiments show promising outcome and open up new avenue for further exploration for protein function annotation.