Pharmaceutical

Achieving a 4-fold conversion increase through ene-reductase discovery and optimization

The challenge

A pharmaceutical company sought a viable biocatalytic route for the stereoselective reduction of a specific substrate using an ene-reductase.

While active enzymes were known, they were heavily patented and commercially inaccessible. The client needed to discover novel, unencumbered starting points from uncharacterized sequence space, and subsequently boost catalytic activity to meet stringent targets for industrial viability.

 

Our solution

After conducting a feasibility assessment that showed positive results and reinforced the project’s viability, we initiated a two-phase campaign to secure the ideal biocatalyst:

 

Phase 1: Enzyme Discovery

We began by virtually screening a massive library of more than 400,000 sequences in silico. Predictive computational filtering and conformational modeling allowed us to navigate high sequence variability and narrow the pool down to just 20 highly promising sequences for experimental testing.

Benchmarked against known patented enzymes, this screen delivered 5 functional hits, securing clear freedom to operate and viable starting points.

 

Phase 2: Enzyme Optimization

With a functional starting enzyme secured, the goal shifted to maximizing its conversion rate. Leveraging our enzyme optimization platform, Zymevolver, we explored millions of in silico designs across both active-site and distal positions.

Operating on an accelerated timeline of just one month, we performed a single round of optimization and selected 140 variants for laboratory testing.

 

Results & impact

  • Patent-free starting points unlocked
    Successfully bypassed existing patent barriers by identifying multiple active, unencumbered wild-type scaffolds from global sequence space, securing long-term freedom to operate and providing viable, proprietary starting points for commercial synthesis.
  • Conversion increase
    The best-performing variant achieved a 4-fold increase in conversion compared to the initial discovery hit, helping to reduce downstream costs and increase API purity.
  • Accelerated timeline
    Delivered both diverse, unencumbered starting points and highly active, industrially viable ene-reductase variants in just one month and with a single computational round, proving the efficacy of combining physics-AI methods for in silico discovery and optimization.

 

Performance Highlights

4-fold
Conversion increase
The best optimized variant delivered a 4-fold improvement in conversion compared to the initial baseline hit. 
1
Month
Only one round of computational optimization and a single month were required to test 140 variants and secure 8 improved hits.
Page background

Request a sample