Antimicrobial peptides (AMPs) are naturally produced by pro- and eukaryotes and are promising alternatives to antibiotics to fight multidrug-resistant microorganisms. However, despite thousands of AMP entries in respective databases, predictions about their structure-activity relationships are still limited. Similarly, common or dissimilar properties of AMPs that have evolved in different taxonomic groups are nearly unknown. We leveraged data entries for 10,987 peptides currently listed in the three antimicrobial peptide databases APD, DRAMP and DBAASP to aid structure-activity predictions. However, this number reduced to 3,828 AMPs that we could use for computational analyses, due to our stringent quality control criteria. The analysis uncovered a strong bias towards AMPs isolated from amphibians (1,391), whereas only 35 AMPs originate from fungi (0.9%), hindering evolutionary analyses on the origin and phylogenetic relationship of AMPs. The majority (62%) of the 3,828 AMPs consists of less than 40 amino acids but with a molecular weight higher than 2.5���kDa, has a net positive charge and shares a hydrophobic character. They are enriched in glycine, lysine and cysteine but are depleted in glutamate, aspartate and methionine when compared with a peptide set of the same size randomly selected from the UniProt database. The AMPs that deviate from this pattern (38%) can be found in different taxonomic groups, in particular in Gram-negative bacteria. Remarkably, the ��-core motif claimed so far as a unifying structural signature in cysteine-stabilised AMPs is absent in nearly 90% of the peptides, questioning its relevance as a prerequisite for antimicrobial activity. The disclosure of AMPs pattern and their variation in producing organism groups extends our knowledge of the structural diversity of AMPs and will assist future peptide screens in unexplored microorganisms. Structural design of peptide antibiotic drugs will benefit using natural AMPs as lead compounds. However, a reliable and statistically balanced database is missing which leads to a large knowledge gap in the AMP field. Thus, thorough evaluation of the available data, mitigation of biases and standardised experimental setups need to be implemented to leverage the full potential of AMPs for drug development programmes in the clinics and agriculture.
