Understanding the metabolic potential of organisms or a bacterial community based on their (meta) genome requires the reliable prediction of an enzyme's function from its amino acid sequence. Besides a remarkable development in prediction algorithms, the substrate scope of sequences with low identity to well-characterized enzymes remains often very elusive. From a recently conducted structure function analysis study of PLP-dependent enzymes, we identified a putative transaminase from Bacillus anthracis (Ban-TA) with the crystal structure 3N5M (deposited in the protein data bank in 2011, but not yet published). The active site residues of Ban-TA differ from those in related (class III) transaminases, which thereby have prevented function predictions. By investigating 50 substrate combinations its amine and omega-amino acid:pyruvate transaminase activity was revealed. Even though Ban-TA showed a relatively narrow amine substrate scope within the tested substrates, it accepts 2-propylamine, which is a prerequisite for industrial asymmetric amine synthesis. Structural information implied that the so-called dual substrate recognition of chemically different substrates (i.e. amines and amino acids) differs from that in formerly known enzymes. It lacks the normally conserved 'flipping' arginine, which enables dual substrate recognition by its side chain flexibility in other omega-amino acid:pyruvate transaminases. Molecular dynamics studies suggested that another arginine (R162) binds omega-amino acids in Ban-TA, but no side chain movements are required for amine and amino acid binding. These results, supported by mutagenesis studies, provide functional insights for the B. anthracis enzyme, enable function predictions of related proteins, and broadened the knowledge regarding omega-amino acid and amine converting transaminases.