are also using the power of machine learning and AI to design novel enzymes or improve existing ones. They analyze gene sequences from public and proprietary databases and use predictive models to “write” new enzyme sequences.
If you know anything about models, you know the axiom that the model is only as good as the dataset it has been trained on. This is where Ginkgo Bioworks has a massive advantage., which the company refers to as their “biological portfolio”, contains DNA sequences that can be used to train the machine learning algorithms to recognize the features that contain the desired qualities for enzyme engineering: “Because we run our labs like factories,” says Kelly.
Ginkgo says its database currently contains close to 1 billion sequences, and all that wealth of genetic information is available to their customers: “It's just a service contract away from anybody that wants to develop a product,” says Kelly.Biology is incredibly complex, so every model-generated sequence still needs to be tested in the lab. The Codebase is just “the fuel for the design engine”.
“What we really want to drive home to our customers is that you don’t need to do it yourself – we have invested half a billion dollars into infrastructure and automation so you don’t have to,” says Kelly. ‘We succeed when our customers succeed’ is the philosophy Ginkgo is banking on.