Nature serves as a rich source of inspiration. We've ingeniously harnessed enzymes and nucleic acids, transforming them into powerful tools like genome editors and PCR diagnostics. Nevertheless, nature's abundant offerings have sometimes confined us within the 'resource curse' paradigm, wherein biology has predominantly involved tweaking existing molecules. To unlock the full potential of biotechnology, enabling solutions to pressing challenges like aging, climate change, and sustainable energy, we must liberate ourselves from this cycle. The key lies in embracing Generative AI and Generative Biology.
Generative Biology
Projects
Jorna Therapeutics works on RNA editing. With SkyEngine de novo-designed RNA search engine, CasNova, its NanoBE and NanoPE systems are revolutionizing what's possible with RNA manipulation.
RNA Editing
Kana works on genome editing, and it is currently in stealth mode.
Kana Medicine
Genome Writing
Entrega aims to build the next generation delivery systems for nucleic acid medicines. Its VAILs (vehicles by AI layering) are a new class of delivery tools created by our generative AI.
Delivery
We are working on the first phase of Quantum Sky, and would love to hear from you! If you are interested in our updates, please hit the "Subscribe" button below. If you want to reach out and chat, please send us a note through the contact form. We look forward to hearing from you.
Contact+Subscribe
Enter, SkyEngine
SkyEngine, our Generative AI engine, operates on both proteins and RNAs. The biomachines it generates have translated into tangible products, making a real-world impact. For instance, it has facilitated the creation of de novo-designed CRISPR-Cas proteins, aiding our clients in launching COVID-19 diagnostic kits and RNA editing pipelines.
SkyEngine implements a hybrid model of Large Language Model (LLM) and Geometric Deep Learning (GDL), this is the number of parameters in the foundation AI model of SkyEngine.
6.5B
This is the number of sequences and structures the model was trained on.
1.2B
Compared with the state-of-the-art models, such as AlphaFold2, our model achieves 100X acceleration while producing similar or even better predictions.
100X
It only takes a few examples for the SkyEngine to generate working de novo proteins
5