Why Elix Discovery™
01.
Data Advantage:
16 Pharma Partners
AI models trained on proprietary data from 16 pharma partners, covering activity, ADME, and toxicity.
02.
Novel Structures
Beyond Human Imagination
Designs novel, non-obvious structures, proven across multiple programs.
03.
PhD-Led Consulting & Support
Experienced PhD scientists provide hands-on onboarding, workflow integration, and ongoing support.
Features
Predictive Models
・Predict molecular profiles, including activity, ADMET, and physicochemical properties.
・Support from traditional ML models to cutting-edge DL models.
・Automatically build the best model for chemists.
Generative Models
・Generate molecules with desired properties, such as activity and ADMET.
・Support both ligand-based (LBDD) and structure-based (SBDD) drug design.
・Handle multiple scenarios—de novo generation, lead optimization, linker generation, etc.
Integration of physics-based simulation and AI
・Each can take advantage of complementary strengths, and the combination can help deal with a lack of data.
・Optimize structures by docking simulation with GPU acceleration.
・Docking simulation can be used even in the absence of activity data.
・Predictive models and pharmacophore model integration are also possible.
FAQ
- Are medicinal chemists your main target user?
- Yes. Elix Discovery™ is developed with the concept that it should be intuitive and easy to use for medicinal chemist; not only for computational scientists.
- Will Elix Discovery™ replace medicinal chemists?
- No. To obtain desired outcomes using AI drug discovery tools, iterative modeling cycles where medicinal chemists review results and fine-tune AI models. Elix Discovery™’s strength is enabling medicinal chemists to run these iterations intuitively and efficiently.
- What types of data can I use to train models in Elix Discovery™?
- Elix Discovery™ allows you to train models using your proprietary in-house data. Depending on your use case, you can either fine-tune an existing pre-trained model with your data or train a model from scratch. Depending on your plan, you can also use models trained on data from 16 pharmaceutical companies (currently accepting applications for the waiting list).