Chemistry & ADMET · AstraZeneca
REINVENT 4
Generates and optimises small molecules against multi-objective property profiles using reinforcement learning.
What REINVENT 4 is
REINVENT 4 is a generative chemistry model developed by AstraZeneca for de novo small-molecule design and optimisation.
The tool addresses multi-parameter drug design challenges by directing molecular generation toward composite targets, balancing properties such as potency, selectivity, ADMET (absorption, distribution, metabolism, excretion, and toxicity), and synthesizability.
How it works
The system uses reinforcement learning over a SMILES generator to explore chemical space. Candidate structures are evaluated against a user-defined multi-objective scoring function to guide the generation process toward molecules meeting specified criteria.
Key facts
- Version
- 4.0
- Domain
- Chemistry
- Modality
- Small molecule
- Supported inputs
- Scaffold / seed SMILES, objective definition, number of molecules
- Output format
- Table of generated molecules with objective scores
- Required GPU memory
- 16 GB
What you can use it for
- Multi-objective small-molecule optimisation
- Optimise molecular candidates against a combined profile of potency, selectivity, ADMET traits, and synthetic viability.
- Scaffold-based de novo design
- Generate novel compound structures derived from a provided seed or scaffold SMILES input.
- Targeted chemical library generation
- Sample a user-defined number of molecules scored against specified design objectives.
Run it here
No installation and no GPU of your own: fill in the inputs and it runs on our servers. Results stay in your browser unless you sign in.
Not runnable yet
This model isn't available to run right now.Run this model
REINVENT 4
Strengths and limits
Good at
- Guides molecular search directly against composite, multi-objective scoring functions.
- Accommodates scaffold-directed exploration via seed SMILES strings.
Watch out for
- Requires a GPU with at least 16 GB of memory to run.
- Dependent on the formulation of an objective function to evaluate and steer molecular properties.
Inputs and outputs
You provide
Scaffold / seed SMILESoptional
Objectiveoptional
Comma-separated scoring components.
Moleculesoptional
You get back
Molecules with objective scores.
Licence
Commercial use permitted · Apache-2.0
The Apache-2.0 licence allows use in commercial research and products.
Questions
Can I use REINVENT 4 commercially?
Commercial use permitted. The Apache-2.0 licence allows use in commercial research and products.
How do I run REINVENT 4 without installing anything?
Use the form under “Run it here” on this page: fill in the inputs and press Run model. It runs on our servers on 2× A100, typically in 10min, and a run costs about 3,500 credits. Visitors without an account get 1,000 credits a month.
What inputs does the model take?
It takes an optional scaffold or seed molecule in SMILES format, a defined objective, and the number of molecules to generate.
What are the model outputs?
The model outputs a table listing the generated molecules alongside their calculated objective scores.
Which molecular properties can be targeted during design?
The composite objective function can target properties including potency, selectivity, ADMET characteristics, and synthesizability.
Written from REINVENT 4's own documentation. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.