A Brief Guide to the APD (Updated Jan 2026)
1. Database scope
Antimicrobial peptides (AMPs) or host defense peptides (HDPs) are essential components of innate immune systems to protect host from infection and are promising candiates for developing new antibiotics. To promote research, education, and information retrieval and knowledge discovery in the antimicrobial peptide field, we initially created the Antimicrobial Peptide Database and data analysis system (APD) in 2003. The peptides stored in this database were gleaned from the literature (PubMed, PDB, Google Scholar, UniProt, and AI-assisted web search) manually in the past 23 years. While the APD has facilitated peptide design and activity prediction by machine learning, the APD6 has deployed an unprecedented AMP information pipeline (AMPIP) for developing advanced AI prediction models.
For scientific rigor, the APD3 first established a set of criteria for peptide registration in 2011.
The APD6 has expanded the collection of synthetic peptides and added AI predicted AMPs to enrich the AMPIP. Hence, peptides will be registered into this database if
(1). they are "natural AMPs", "predicted" or "synthetic" peptides;
(2). antimicrobial activities are demonstrated (MIC <100 uM or 100 ug/ml);
(3). the amino acid sequences of the peptides have been elucidated, at least partially; and
(4). antimicrobial peptides contain less than 100 amino acid (aa) residues (Starting October 2012, the APD database also included some small yet important antimicrobial proteins with less than 200 aa).
To increase sequence uniqueness, peptides from different species with the same amino acid sequence occupy the same entry (135 peptides, search using "found in multiple species" in the Additional Info field, enabled in 2012). Likewise, amidated and non-amidated sequences are also placed in the same entry and described under SAR. Our practice led to an authentic data set for natural AMPs first reported in the APD3. When one refers to over 3000 natural AMPs (recently unified and refined), this database is the original source for this info. Hence, the APD contains a core data set for antimicrobial peptide design and prediction (Wang, 2023).
2. Database capabilities and AMP sub-databases
Our goal is to develop the antimicrobial peptide database into the most comprehensive tool for discovery timeline, naming (nomenclature), classification, structure, function, information search, statistics, prediction, and design of antimicrobial peptides. As an online e-Dictionary, you can search the database
for the unprecedented detailed info for any peptide (using the standard 20 aa) in many ways as listed below:
(1). Peptide full sequence, partial sequence, or motifs in the single-letter amino acid code such as DP and ILIKEAPD;
(2). In the name field, peptide name such as magainin, thanatin, gomesin, kalata B1, cecropin, aurein 1.2, dermaceptin;
(3). Peptide family name such as defensin, cathelicidin, and histatin;
(4). Common name for peptide source species such as human, cattle, frog, fish, spider, snake, and fungii (actual search words);
(5). peptide sources: (a) "natural AMPs", e.g. three domains (bacteria, archaea, eukaryotes) or six kingdoms (bacteria, archaea, protists, fungii, plants, and animals); (b) "predicted" (e.g., machine learning); and (c) "synthetic" peptides. Quoted words are searchable.
(6). Peptide post-translational modification using search keys such as XXA (amidation), XXC (cyclization), XXD (D-amino acids), XXE (N-terminal acetylation), XXG (glycosylation), XXH (halogenation), XXK (hydroxylation), XXO (oxidation), XXP (phosphorylation), XXQ (N-terminal cyclic glutamate), and XXS (sulfation); A more complete list can be downloaded here.
(7). Still in the name field, AMPs that form dimers can be searched by entering "dimer" (hetero or homo);
(8). Peptide-binding partners or targets using search keys such as BBS (carbohydrates), BBL (lipopolysaccharides, LPS, endotoxin) and BBII (cations such as Zn2+), BBN (nucleic acids), BBMm (biological membranes) (listed in the Glossary);
(9). Peptide original location: PDB, UniProt or
Reference;
(10). Cationic, neutral, or anionic AMPs based on net charge: <0, =0 or >0;
(11). Hydrophobic residue content (Pho%);
(12). Peptide length (size);
(13). Biophysical methods for structure determination: NMR spectroscopy, X-ray crystallography, or Circular Dichroism (CD);
(14). Structural classification such as alpha-helix, beta-sheet, alphabeta, or non-alphabeta;
(15). Contributing authors or year of AMP publication in the references;
(16). In the biological activity field, antibacterial (Gram-positive only or Gram-negative only), antiviral, antifungal, and anticancer peptides. It appears that AMPs have multiple functions. The APD is the first database recorded such functions in each version and will continue to annotate new functions with time;
(17). Source species search in scientific name (e.g. Homo sapiens). This allows for the search of a collection of AMPs from a specific living species so far collected into the APD.
(18). In 2020, the APD also first enabled "microbiota" search in the Source Organism field;
(19). Antimicrobial activity against any species can be searched in the Additional Information field by entering "E. coli", "S. aureus", or "C. albicans"; Note that E.coli (searchable) is not a typo but indicates that "this E. coli strain was not inhibited by the peptide with MIC > 100 ug/ml or 100 uM". In 2023, the APD has established a full activity annotation system by creating a third "uncertain" categoty, e.g., E-coli (searchable), for any observed activity not greater than 100 uM (e.g., >16 uM) (read the paper here).
(20). A more general list of AMPs with synergistic effects can be searched in the Additional Info field by using "synergy" (synergy:AMPs: 49 between AMPs; synergy:antibiotics: between AMPs and antibiotics);
(21). Also antimicrobial robustness info: format: 75 NaCl-sensitive:E. coli or 46 MgCl2-insensitive:S. aureus peptides can be searched in the Add Info field; pH-sensitive (see the APD6 for more)
(22). Resistance development:yes or Resistance development:no in the Additional info field;
(23). MOA (mechanism of action) in the Add info field (search format: MOA:bacteria; MOA:fungi);
(24). Recombinant production in the Add info field (search format: Recombinant production:bacteria; Recombinant production:yeast);
(25). In vivo toxicity in the Add info field;
(26). Animal model (in vivo efficacy search format in the Add info field: animal model:mouse; animal model:fish); and
(27). Any combination of two. See the APD6 paper for more search options.
The above annotated information makes the APD most comprehensive for each peptide entry (see LL-37 as an example).
Kingdom search: the APD is the first database that adopted the six life kingdom classification initiated in the APD2 and completed in the APD3. You can search AMPs from a defined life kingdom by entering bacteria, archaea, fungi (search with fungii), plants, and animals in the NAME field. For example, a subdatabase for bacterial AMPs can be generated by entering the word "bacteria" into one of the search boxes in the name field, and any properties of the bacterial AMPs (i.e. bacteriocins) can be searched in the usual way. Likewise, users can search AMPs from plants or animals (amphibians, fish, reptiles, birds, invertebrates, insects, spiders, mollusks, worms, crustacean, spiders, etc.).
As one of the recent addition, the APD has also annotated AMPs from microbiota. This information can be searched via "Source" (e.g., human microbiota:gut).
The database also has the Prediction and Design Interfaces from the very beginning since 2003. The APD database programmed the first calculation and prediction interface in 2003. After you input the sequence, the program will calculate select properties of the peptide (e.g. net charge, length, hydrophobic residue% and amino acid composition). If the calculated parameters are out of the APD-defined space for natural AMPs, the program will stop (Nov 2013 version). Otherwise, the system will traverses the database and does pairwise sequence alignment. The APD will then list many peptide sequences that are similar to your input (initially limited to 5 most similar sequences). The program will calculate a similarity score and displays the differences between the input and database sequences. You can improve the activity of the peptide you designed based on the alignment results. Hence, it is useful to compare your sequences with the entries in the APD before you report your new AMPs.
The APD is also the first in providing statistical information on peptide sequence, structure and function of all entries or a group of peptide entries of similar properties such as anticancer or from the same sources such as bacteria. Examples of statistic data about AMPs can be viewed here.
For the definition of selected terms or abbreviations related to AMPs, please go to Glossary.
For AMP papers contributed by the Wang laboratory, please go to Contact and Publications.
In 2023, an international symposium was held to celebrate the 20 anniversary of the APD. You can listen to the invited lectures by visiting Conference, which also contains a list of the 2026 peptide conferences.
For frequently asked questions from users, please refer to the FAQ page.
Selected peptide parameters and properties can be calculated using the peptide calculator of the APD. In addition, the APD has also generated links to Tools for users to calculate other properties of newly discovered peptides.
The APD also first set up links to existing databases in 2003 (also listed in a tool review in 2007). This link has now been extensively updated in the APD6 and can be accessed here.
3. Database history, update and further development
Rome was not built in a day. The Antimicrobial Peptide Database (APD published in 2004) was originally created by a graduate student, Zhe Wang, as his master's thesis in the laboratory of Dr. Guangshun Wang. The project was initiated in 2002 and the first version of the database was open to the public in August 2003. It contained 525 peptide entries, which can be searched in multiple ways, including APD ID, peptide name, amino acid sequence, peptide length, net charge, hydrophobic content, antibacterial, antifungal, antiviral, anticancer, hemolytic activity, original location, PDB ID, 3D structure, and methods for structural determination. The database also programmed the first interfaces for peptide prediction and design. Peptide statistics for all, a group of, or a single of peptide can also be calculated.
The 2nd version of our database reported 1,228 antimicrobial peptide entries. Major features added are classification of AMPs based on peptide sources, families, chemical modifications, and binding targets. A summary of the new developments and database findings, including a demo of database-aided peptide design, is described in APD2 published in 2009. The new web design is credited to Sophie Wang.
The 3rd version (APD3 published in 2016) reported 2,619 peptides. New educational web pages were created for FAQs, interesting AMP discovery timeline, classification, nomenclature, AMP facts, additional tools, Sequence downloads, and APD News (What's New). A unified peptide classification scheme has been introduced and updated. Peptide registration criteria were proposed so that this model database focuses on natural peptides and generates a needed core data set to decipher the design principles of natural antimicrobial peptides. Peptide biological sources were finally classified into six life kingdoms. In addition, the prediction interface has been improved and more peptide properties can be calculated.
In 2020, a Protein Science paper was published to describe new peptide functions and the increase in peptide entries. The design principles of natural AMPs in different kingdoms or classes were highlighted. This version is now referred to as the APD4 published in 2020 (i.e., APD2020) with 3081 entries.
In 2022, additional peptide functions and peptide entries from microbiota were added along with a description of database evolution in the past 18 years. Also, the database was reprogrammed to enhance cybersecurity. This paper is now referred to as the APD5 published in 2022 (i.e., APD2022) with 3273 entries.
In 2023, an international symposium was held to celebrate the 20 anniversary of the APD. You can listen to the invited lectures by visiting the APD Conference.
In 2025, the APD6 reported a total of 5188 peptides (3306 natural AMPs, 1380 synthetic AMPs, and 239 predicted AMPs with activity validated and sequence known). The novel AMP information pipeline (AMPIP) has substantially enriched the information scope for antimicrobial peptides.
The APD is regularly updated and formatted. As of Sept 10, 2025, this database contained a total of 5680 peptides and proteins, including 3351 natural, 1733 synthetic, and 329 predicted AMPs.
We appreciate users who emailed us their depositions, suggestions, and corrections. We were able to incorporate them in the current version and give credits by including the provider's name in the related entries (e.g. T. Stein, Erik Martin, Adel Ghorani-Azam, Hadi Zare-Zardini, M. Bassam Alkotaini, Brice Felden, Sven-Ulrik Gorr, Jun Wang, and Manjul Lata). Please contact us if your newly discovered AMPs escaped our attention.
4.Database copyrights and important legal notice
The APD adopted the open access model from the beginning to promote research,education, and development. It also facilitated others to build competing databases by directly copying the APD without permission. As we posted in 2003, the antimicrobial peptide database is copyright-protected. This includes the phrase "Antimicrobial Peptide Database", which is the specific, copyrighted name for the APD. Use of this specific phrase to refer to a different database is a violation.
Disclaimer: Although every effort is made to make the database as complete and accurate as possible, we do not assume liability for any claim due to the use of the APD and its derivatives.