How to Use the BRENDA Enzyme Database: From EC Classification to Industrial Enzyme Selection
Published on November 4, 2026

Figure 1 | BRENDA enzyme database: hub of enzyme functional data
【Category】:D (AI / algorithms / platform) · cross-category A (enzymes / proteins)
In biocatalysis, pharmaceutical metabolism, and synthetic biology, “picking the right enzyme” is often the difference between success and failure. Yet the world of enzymes is enormous: different enzymes catalyze different reactions, and the same class of enzyme can behave very differently across organisms and tissues in terms of kinetics and stability. The BRENDA enzyme database acts like an authoritative “directory of enzyme profiles” — it concentrates manually curated enzyme data from the literature so you can quickly find out what an enzyme catalyzes, how fast it works, and what conditions it prefers, saving you from trial and error in real enzyme engineering and industrial production.
Background: Why do research and industry both need an “enzyme dictionary”?
Takeaway:Founded in 1987 in Braunschweig, Germany, the BRENDA enzyme database has remained the world’s leading source of enzyme functional data; by turning scattered literature information into searchable, comparable structured data, it has become the common starting point for research and industrial enzyme selection.
BRENDA stands for BRaunschweig ENzyme DAtabase. It did not start as a website — it was originally published as a series of enzyme handbooks; as the data grew, it became a public online relational database in the late 1990s and has been continuously curated at the Technische Universität Braunschweig since 2007. Today it is maintained by the Leibniz Institute DSMZ in Germany and is both an ELIXIR Core Data Resource and a Global Core Biodata Resource, widely regarded as critical data infrastructure for the international life-sciences community.
Its core value comes down to three points:
· Manual curation:Most data are extracted, checked, and annotated item by item from the primary literature rather than machine-scraped, so quality and traceability are high (each record carries its source and PubMed link).
· Structured search:Enzyme data are organized under a single EC classification system, so you can browse by reaction type or filter by substrate, biological source, and parameter ranges.
· Open access:The online version is free under CC BY 4.0; both academic and industrial users can access it as needed, and can integrate it into their own pipelines via downloads and APIs.
In short, BRENDA solves the pain point that “enzyme information is scattered across thousands of papers and hard to compare side by side,” giving the enzyme field a shared, searchable, comparable, downloadable “dictionary.”
What is BRENDA, and what does its “data structure” look like?
Takeaway:BRENDA organizes data by the IUBMB Enzyme Commission classification; each EC number corresponds to a class of enzymes that catalyze the same chemical reaction, and it is then enriched with fields such as reaction, substrates/products, kinetics, cofactors, inhibitors, optimal conditions, biological source, and stability — a tree of “one number, many attributes.”
To understand BRENDA, first understand the EC number. The EC (Enzyme Commission) number is the enzyme classification system established by IUBMB. It classifies enzymes not by their source but by the chemical nature of the reaction they catalyze. One EC number represents “enzymes that catalyze the same class of reaction,” so EC 3.4.21.4 may come from bovine, porcine, or human pancreas, yet all of them hydrolyze the same type of peptide bond.
The EC number has a four-level hierarchy, the first digit indicating one of seven main classes:
· EC 1 oxidoreductases
· EC 2 transferases
· EC 3 hydrolases
· EC 4 lyases
· EC 5 isomerases
· EC 6 ligases
· EC 7 translocases
The IUBMB Enzyme Commission has defined more than 8,300 EC numbers, and BRENDA covers all of them (including obsolete ones). Under each EC number, BRENDA attaches a set of multidimensional attribute fields, mainly including:
· Reaction and specificity:catalyzed reaction, substrates/products, specificity (whether it acts on a particular bond or structure).
· Functional parameters:Km (Michaelis constant, reflecting enzyme–substrate affinity), Vmax (maximum reaction rate), kcat (turnover number), kcat/Km (apparent catalytic efficiency), and more.
· Cofactors and ligands:natural or in vitro substrates/products, inhibitors, activators, cofactors, bound metals.
· Occurrence and source:organism, tissue, cellular localisation.
· Optimal conditions:optimal pH, optimal temperature, oxidation / organic-solvent tolerance ranges.
· Structure and stability:molecular weight, structural information, stability (storage, temperature range, oxidation stability).
· Application and engineering:application scenarios, mutant and engineered-variant properties, isolation and preparation, related diseases.

Figure 2 | BRENDA data model: multidimensional fields under an EC number
This “one number, many attributes” tree lets researchers and engineers compare enzymes of different origins and conditions under the same logic.
How to search and select enzymes with BRENDA?
Takeaway:To select an enzyme with BRENDA, first locate the EC number by substrate or reaction, then filter step by step by biological source, catalytic efficiency (kcat/Km), optimal pH/temperature, and tolerance — narrowing candidates from “usable” to “useful.”
BRENDA offers a range of search entry points; choose as needed:
· Full-text Search:type in enzyme name, substrate, product, or literature keywords to locate relevant entries quickly.
· Advanced Search:combine multiple fields (e.g., “a given EC number + a given organism + a given Km range”) for precise filtering and drilling down.
· Ligand Structure Search:draw or enter a substrate/product structure to find enzymes that catalyze its transformation — especially useful when “I have a class of substrates and want enzymes that act on them.”
· EC Explorer and TaxTree:browse top-down by the EC tree or the taxonomic tree to survey the enzyme landscape of a reaction class.
· Gene Search and structure search:find specific genes or link known enzyme structures (CATH, SCOPe).
· Functional Parameter Statistics:view distributions of a parameter class to judge “typical values” versus “outliers.”
The selection workflow can be:
1. Define the reaction:clarify the chemical reaction or substrate to be catalyzed.
2. Gather candidates:use the substrate/structure or reaction to reach the EC number and collect the candidate enzymes “catalyzing that reaction.”
3. Filter conditions:filter step by step by biological source (which may affect expressibility, side reactions), catalytic efficiency, optimal pH/temperature, and tolerance.
4. Compare head vs. tail:use parameter statistics to see where a candidate sits among the class and identify “more desirable” ones.
5. Check against production:cross-validate candidate activity and stability with host expression and process conditions.
For automation, BRENDA also offers programmatic interfaces such as SOAP and SPARQL, as well as SBML output and full downloads, making it easy to feed enzyme data into internal databases or computational pipelines (its new website and API are also being deployed).
Process conditions and enzyme engineering: from data to scale-up
Takeaway:The Km, Vmax, kcat, and optimal pH/temperature data in BRENDA are the basis for designing reactor processes and quality control; feeding these measured values back into an enzyme-engineering platform lets you identify thermal-stability, tolerance, and activity bottlenecks before scale-up.
An enzyme that is “usable” in the lab is not necessarily “useful” in a factory. During scale-up, what matters is:
· Catalytic efficiency:kcat/Km drives the yield per unit of enzyme; Km magnitude influences substrate-concentration design.
· Optimal conditions:optimal pH and temperature define the reactor operating window; deviating from them reduces both activity and lifetime.
· Tolerance:tolerance to organic solvents, oxidation, high temperature, and high substrate concentration — BRENDA records these stability data, key to judging whether an enzyme can withstand industrial conditions.
· Substrate specificity:whether it catalyzes side reactions affects product purity and downstream separation cost.
So a mature workflow is often: use databases such as BRENDA to lock in candidate enzymes and process windows → characterize the real parameters in the lab → use an engineering platform to improve bottlenecks → return to the data table to verify improvements. That way, the quality-control parameters (pH, temperature, substrate concentration, time) are grounded from the start rather than found by trial and error.

Figure 3 | Application loop: from BRENDA enzyme selection to process optimization and engineering
From BRENDA to MatwingsVenus®(protein agent): an AI-driven enzyme-engineering loop
Takeaway:MatwingsVenus®(晓鹜®) uses enzyme functional data from databases such as BRENDA as its starting point, then combines AI structure and sequence prediction, protein design, and directed evolution to form a “discover–characterize–engineer–re-validate” loop that frees people from sifting through vast literature.
Databases answer “which enzymes exist and what are their parameters,” but making an enzyme “better” usually requires rounds of prediction and engineering. That is exactly the focus of MatwingsVenus®(晓鹜®) — the AI-driven enzyme-engineering platform of Matwings Technology:
· Data as the starting point:use the kinetics, stability, and reaction information from authoritative databases such as BRENDA to quickly build a candidate-enzyme profile and avoid starting from scratch in the literature.
· AI structure prediction:predict three-dimensional structure from sequence and assess active-site and stability-critical residues.
· Protein design:within a design-model / directed-evolution framework, propose candidate mutations and predict their effect on activity and stability.
· Loop validation:feed measured post-engineering parameters back into the data tables, forming a “data–prediction–experiment–data” flywheel that progressively approaches the performance industrial processes need.
In other words: BRENDA provides a structured map of the “known world,” and MatwingsVenus®(晓鹜®) uses AI to explore the boundary of “more useful enzymes.” The two complement each other, upgrading “find enzymes with data” to “engineer enzymes with data + AI.”
FAQ
Is the BRENDA enzyme database free?
The online version is free under the CC BY 4.0 license. You can search, download, and call its interfaces for free as long as you cite the source; commercial in-house deployments are available through its licensing distribution channel. Please refer to the official terms & conditions for full details.
How does BRENDA classify enzymes?
By the IUBMB Enzyme Commission (EC) numbers. EC numbers classify enzymes by the chemical nature of the reaction they catalyze (not by source or sequence), using a four-digit hierarchy; the first digit corresponds to one of seven main classes (oxidoreductases, transferases, hydrolases, lyases, isomerases, ligases, translocases).
What enzyme parameters can I find in BRENDA?
You can find catalyzed reactions, substrates/products, specificity, kinetic parameters (Km, Vmax, kcat, kcat/Km), cofactors, inhibitors, organism and tissue source, cellular localisation, optimal pH and temperature, oxidation/solvent tolerance, molecular weight and structure, stability, applications, mutant and engineered-variant properties, and the literature source of each record.
Can BRENDA directly engineer an enzyme for me?
No. BRENDA is a data repository that provides authoritative enzyme information and selection rationale, but it does not do sequence design or directed evolution. To “engineer an enzyme,” you need to build on data such as BRENDA and then use protein design and directed evolution tools (such as the MatwingsVenus®(晓鹜®) platform).
Summary
Through the unified IUBMB EC classification, manually curated literature data, multidimensional enzyme-function fields, and open access, the BRENDA enzyme database has become a trusted “central dictionary” for the enzyme field. It lets researchers and engineers quickly locate enzymes that catalyze a given reaction, compare their kinetics and stability, and assess whether they fit specific process conditions. When “finding candidate enzymes” is not enough, feeding BRENDA’s structured data into an AI enzyme-engineering platform such as MatwingsVenus®(晓鹜®) can push “find enzymes with data” forward to “engineer enzymes with data + AI,” providing closed-loop support for performance gains from lab to factory.
【One-sentence Takeaway】:BRENDA is the authoritative central library of enzyme functional data; combined with an AI enzyme-engineering platform, it goes straight from EC classification to industrial enzyme selection and directed engineering.
