SMALT + AI directed evolution
Custom Enzyme Development Services
- Custom R&D service, not a catalog product
- SMALT + AI platform with partner Avolution
- Academic base: Sun Yat-sen University
Tianyi Health Tech offers custom enzyme development for food, industrial and pharma customers: we connect your project to a SMALT + AI directed-evolution platform built with our partner Avolution, covering screening, optimization, expression and application validation. This is a project-based service, so tell us your target and we will scope the work.
At a glance
Key Facts
Overview
Platform Positioning
Customers need an enzyme that works under their real formulation, process and cost constraints, not only a catalog enzyme. With Avolution, Tianyi uses SMALT targeted mutagenesis and AI closed loops (the SSA evolutionary accelerator) to address low mutation efficiency, slow screening and the disconnect between wet-lab and computational work.
The platform’s IP covers SMALT and AI protein-design methods, with background IP held under exclusive commercialization pathways from Sun Yat-sen University.
Pain points
Pain Points and Solutions
Inefficient mutation
Classic in vitro single-site mutation is shallow. In vivo multi-protein, high-frequency targeted mutagenesis explores much deeper.
Slow screening
Low throughput and 4–6 month cycles are typical. Microfluidic single-cell screening at 109 scale can compress this to 1–2 months.
Missing data loops
Genotype-phenotype data from each round feeds multi-objective AI optimization that keeps learning and raises hit rates.
Application fit
Variants are checked against your formulation and process, not only a lab assay.
Specification
Platform and Case Data
| Parameter | Specification | Notes |
|---|---|---|
| Service | Custom enzyme screening, optimization, expression and application validation | |
| Platform | SMALT targeted mutagenesis + AI closed loop (SSA accelerator) | Partner: Avolution |
| Screening scale | ≥ 109 by microfluidic / single-cell methods | |
| Typical cycle | Can compress from 4–6 months to 1–2 months | Project-dependent |
| Selected case data (partner-reported) | PETase about 3× mature industrial enzyme activity after two rounds; PlHpaB about 5× activity in one round; methyltransferase about 4.4× relative activity; formate dehydrogenase about 3× peak activity | Results depend on target and project |
| Delivery | Screen, optimize, express, then formulate and validate |
How a project works
From Target to Validated Enzyme
- Define the targetYour substrate, process conditions, cost and performance goals.
- ScreenTargeted mutagenesis and high-throughput single-cell screening at 109 scale.
- OptimizeAI closed loop learns from each round to raise hit rates.
- ExpressExpression and scale-up of the best variants.
- ValidateFormulation and application testing under real conditions.
Applications
Where It Applies
- Food: processing enzymes, protein engineering, fermentation strains, safety and preservation
- Industry: industrial enzymes, biofuels, biomaterials and chemicals, plastics recycling
- Pharma: nanobodies, green intermediate synthesis, cell-therapy modules
Quality
Team & Partnership
The Avolution enzyme core team includes Prof. Shu Wensheng (enzyme platform chief founding scientist), Prof. Yang Yuedong (chief scientist; Sun Yat-sen University and National Supercomputer Center in Guangzhou; AI protein design), and advisors Prof. Liu Jianzhong (head of the SYSU synthetic biology and biocatalysis team) and Assoc. Prof. Liu Li (SMALT mutagenesis). Tianyi connects customer scenarios in food, health and industry to the platform. Full profiles are on the Chinese homepage research-team section.
More ingredients
Related Ingredients
Request a quote
Start an Enzyme Project Inquiry
Research and clinical information is provided for R&D and B2B reference only. These statements have not been evaluated by the Food and Drug Administration. Our ingredients are not intended to diagnose, treat, cure or prevent any disease. Final specifications and regulatory status are confirmed per commercial agreement.