1. Based on the biosynthetic machines
The classic view is that antimicrobial peptides are encoded genetically and expressed to protect the host from infection. The APD has adopted a wide view by including both ribosomally and non-ribosomally synthesized peptides with demonstrated antimicrobial activity. This database focuses primarily on gene-coded peptides. It is important to note that some non-ribosome synthesized peptide antibiotics in this database (e.g. gramicidins, colistin and daptomycin) are already in clinical use. Such information was first entered into the APD and can be searched by entering "AMPs in use" in the additional info field.
2. Based on biological sources
There are various source classification schemes in the literature. The APD started to classify antimicrobial peptides into life kingdoms and domains in 2007 (first published in the APD2 in 2009. Antimicrobial peptides in the APD3 are classified into six life kingdoms: bacteria, archaea (prokaryotes), protists, fungi, plants, and animals (eukaryotes). So, you can search AMPs (bacteriocins) from "bacteria", "plants", "fungii", and "animals" in the APD by entering the quoted words into the NAME field, one at a time. Animal AMPs are further classified based on source families: insects, scorpions, spiders, mollusca (invertebrates), crustaceans, amphibians, fish, reptiles, mammals (vertebrates). The major and well-studied AMPs families in the animal kingdom are cathelicidins, defensins, and histatins. Bacteriocins are further classified (see Glossary).
3. Based on biological functions
This database started to record peptide functions/activities in 2003. Some examples are given below. The list constitutes wheel of function of the APD and is copy-right protected. For a full list, please refer to the APD6 paper. and so on.
4. Based on peptide properties
Traditionally, AMPs are classified based on peptide properties such as charge, hydrophobicity, amino acid composition, and length.
Based on aa composition, there are amino acid-rich peptides (defined greater than 25% in the APD). Examples are Trp-rich, His-rich, Pro-rich, Arg-rich, Gly-rich (classic), Leu-rich, Ser-rich, Lys-rich, Asp-rich, and Ala-rich AMPs (less popular). Based on our definition, we did a systematic analysis in our recent article and did not find AMPs rich in methinine, asparagine, and glutamine in the APD (2022) although they exist in proteins.
Based on net charge, there are cationic (net charge > 0: 88%), neutral (net charge = 0: 6%), and anionic peptides (net charge <0: 6%).
Based on hydrophobic/hydrophilic amino acid composition, there are hydrophobic, amphipathic, and hydrophilic peptides based on hydrophobicity. The APD has a small number of entirely hydrophobic (e.g.,
) or entirely hydrophilic peptides (e.g., ), indicating the dominance of the amphipathic peptides.Also, natural AMPs can be arbitrarily classified based on peptide size (number of amino acids, aa): ultra-short (2-10 aa), short (10-24 aa), medium (25-50 aa), and long (50-100 aa). AMPs greater than 100 aa are antimicrobial proteins (e.g., lysozyme, histones, and RNase 7), which can be searched in the APD by entering "antimicrobial protein" in the NAME field. There are 108 antimicrobial proteins in the APD as of Dec 2025.
5. Based on covalent bonding patterns
This universal peptide classification system (UC) categorizes antimicrobial peptides (or peptides in genenal) into four classes (Wang, G, 2015).
(1) Class I (UCLL): linear one-chain peptides (e.g.
) or two linear peptides not connected via a covalent bond (e.g. enterocin L50).(2) Class II (UCSS): sidechain-sidechain linked peptides (e.g. disulfide-containing defensins or thioether bond-containing lantibiotics). A sidechain-sidechain connection can occur within a single peptide chain or between two different peptide chains.
(3) Class III (UCSB): polypeptide chains with a sidechain to backbone connection (e.g. bacterial lassos and fusaricidins).
(4) Class IV (UCBB): circular polypeptides with a peptide bond between the N- and C-termini (i.e., backbone-backbone connection). Circular peptides have been found in bacteria (e.g.
), plants (e.g. cyclotides), and animals (e.g. theta-defensins).For an update of this classification, please refer to Wang G (2022).
6. Based on 3D structure
In the Wang-edited book, AMPs are classified into four families: α, β, αβ, and non-αβ based on the types of secondary structures. The alpha family consists of AMPs with helical structures (e.g. magainins and LL-37). The beta family is composed of AMPs with beta-strands (e.g. human alpha-defensins). While the alphabeta family comprises both helical and beta-strands in the 3D structure (e.g. beta-defensins), the non-alphabeta family contains neither helical nor beta-strands (e.g. indolicidin). These four families of AMP structures are represented in the main page of the APD . The numbers of AMPs with such structures are annotated in the APD and listed in the statistics.
7. Based on molecular targets
AMPs can be broadly classified into two families: cell surface targeting peptides (e.g. nisins and temporins) and intracellular targeting peptides (e.g. Pro-rich peptides). Cell surface-targeting peptides, including both membrane-targeting and non-membrane targeting peptides, can be further classified based on specific targets such as cell wall/carbohydrates, lipids/membranes, and proteins/receptors. Likewise, intracellular targeting AMPs can be further classified based on the specific target molecules (e.g. proteins, DNA, and RNA). A summary of these molecular targets can be found in a review article Antimicrobial Peptides in 2014.
References
(1) Wang, G.(2022) Unifying the classification of antimicrobial peptides in the antimicrobial peptide database Methods in Enzymology 663, 1-18.
(2) Wang, G (2017) "Antimicrobial Peptides: Discovery, Design and Novel Therapeutic Strategies" (2nd version), CABI, England. Version 1, 2010.
(3) Wang, G. (2015) Improved methods for classification, prediction, and design of antimicrobial peptides. Methods Mol. Biol. 1268, 43-66. PubMed.