Pando BioScience uses generative AI to discover and engineer enzymes for pharmaceutical manufacturing and diagnostic applications. Will Cao, Chief Executive Officer and Co-founder of Pando BioScience stated that their ultra-high-throughput screening platform can screen 1,000 times more enzymes than their competitors. The AI mainly assists with optimising multiple enzyme properties simultaneously.
Cao explained that the screening platform is droplet-based. When a different selective pressure is applied to each droplet, this causes the enzyme to release. The droplets are merged before a sequencing round is performed. Data from the high throughput screening is used to train the AI model with in-house data. Multiple enzyme properties are engineered in one round by the AI, and this method performed 22 times better than the non-AI method.
Pando Bioscience has demonstrated its value through collaborations. For example, their contributions reduced one company’s pharma manufacturing process from 11 chemical steps to 1 enzymatic step, resulting in a $24 million revenue increase for the customer. Another collaboration helped a diagnostic customer to make their enzyme more thermally stable and sensitive.
The type of collaboration services on offer fall into two main categories: enzyme discovery and enzyme engineering. Enzyme discovery explores a large diverse genetic landscape while enzyme engineering allows scientists to improve each variant.
Cao said that Pando BioScience adopts a unique tactic that requires finding an enzyme that does not infringing on any existing IP. This is particularly challenging given the highly crowded IP space. To avoid these existing IPs, Cao took advantage of Pando’s large in-house database that includes data from extreme environments. Fortunately, the team Identified IP-free, high-performance enzymes distinct from market leaders and engineered IP-free variants, matching market leaders in high thermal stability, processivity, and reverse transcriptase activity.
Now, Pando Bioscience is developing its own enzyme assets: it owns the IP for its internal asset, Phi29. Cao mentioned that Phi 29 surpasses market leaders in thermostability and activity and has low GC bias across several genomes. The enzymes are stable and transportable without a cold chain, a highly advantageous property in pharmaceutical manufacturing.
