Human health
- Disease biomarkers
- Cancer
- Infectious disease
- Autoimmune disease
- Cardiovascular disease
- Neurological disease
AI · Structural biology · Protein engineering
Designing the next generation of protein binders
We design novel binding proteins for targets across healthcare, diagnostics, agriculture, food, industrial biotechnology and environmental monitoring. From biological target to functional binder — designed computationally, built experimentally.
Approach
Molecular recognition, engineered from first principles.
Biology runs on molecular recognition. Antibodies became the default tool for detecting, capturing and modulating biological molecules — but discovery is slow, expensive, and bounded by what nature happened to evolve.
Altered Protein is building an AI-native approach to designing binders for targets that are difficult, expensive or impractical to address conventionally.
Instead of searching nature for the right binder, design the binder you need.
The design loop
Design. Test. Learn. Redesign.
Generative design, sequence engineering, structure prediction and experimental validation, joined into one continuous discovery loop.
- 01
Define
Start from a biological target — a structure, a sequence, an epitope.
- 02
Design
Generate candidate binding proteins computationally with generative and protein foundation models.
- 03
Predict
Evaluate structure, binding geometry, stability and developability before anything enters the lab.
- 04
Build
Express and experimentally test the most promising candidates.
- 05
Learn
Return every measurement — including the failures — to the design process.
- 06
Evolve
Improve affinity, specificity, stability and manufacturability. Then run it again.
Back to 01. Every validated design becomes information. So does every failure.
The objective is not to predict proteins. It is to design proteins that work.
Design space
A binder is a specification, not a discovery.
Engineered proteins can be designed around what an application actually requires, rather than around what a screen happens to return.
- Affinity
- How strongly should it bind?
- Specificity
- What should it recognise — and what should it ignore?
- Stability
- Can it survive the environment it will actually be used in?
- Size
- How compact can the binding architecture be?
- Manufacturability
- Can it be produced economically, at scale?
- Function
- Does binding produce a useful biological or analytical outcome?
This moves protein engineering from discovery by screening toward design followed by validation.
The engine
An AI-native protein design engine
A computational stack that connects protein generation, sequence design, structural prediction and functional evaluation.
Generative design
Novel protein backbones and binding architectures.
Sequence engineering
Amino-acid sequences optimised for the intended structure and function.
Structure & interaction prediction
Candidate structures and target–binder interfaces, modelled before synthesis.
Developability
The properties that decide whether a designed protein can become a product.
Experimental learning
Laboratory measurement as ground truth for the next generation of designs.
Feeds back into generative design
A compounding engine: each cycle starts better informed than the last.
From digital design to physical biology
AI proposes.
The laboratory decides.
The bottleneck in protein engineering is not generating another sequence. It is closing the loop between computation and reality — so that every experiment makes the next design decision more informed.
Platform
One platform. Many biological problems.
Protein binders are a horizontal technology. The same underlying design capability produces molecular recognition systems for very different problems.
Diagnostics
- Biomarker detection
- Rapid tests
- Point-of-care diagnostics
- Multiplex assays
- Biosensors
Antimicrobial resistance
- Resistance enzymes
- Carbapenemases
- ESBLs
- Pathogen-specific markers
- AMR surveillance
Agriculture
- Plant pathogens
- Pathogen effectors
- Toxins
- Crop disease biomarkers
- Plant health monitoring
Food & nutrition
- Foodborne pathogens
- Mycotoxins
- Allergens
- Contaminants
- Food authenticity
Industrial biotechnology
- Process monitoring
- Enzyme detection
- Bioprocess analytics
- Contaminant detection
- Manufacturing QC
Environment
- Waterborne pathogens
- Biological toxins
- Environmental contaminants
- Wastewater surveillance
- Field biosensing
Research & life sciences
- Protein capture
- Proteomics
- Cell isolation
- Imaging
- Target validation
Different markets. The same fundamental capability: design a protein that recognises what you need.
Why now
Five things changed at once.
- Generative AI
- New protein sequences and structures can be designed computationally.
- Structural biology
- Prediction systems give unprecedented visibility into molecular structure and interaction.
- Protein foundation models
- Large-scale biological data lets models learn representations beyond sequence alone.
- High-throughput experimentation
- More biological designs can be built and tested than ever before.
- Scalable compute
- Design searches that were impractical are increasingly routine.
ComputeDesignBuildTestLearn
And the loop is getting faster.
Partnerships
Bring us a difficult target.
We work with organisations that have challenging biological targets and need new molecular recognition: poorly characterised targets, low-abundance biomarkers, highly specific molecular signatures, difficult protein surfaces, and applications with unusual stability or manufacturing requirements.
Partner with us lab@alteredprotein.com
Let's engineer biology differently.