Our mission is to accelerate scientific discovery by creating an open community around generative models for science 🚀

Technologies like generative models need to be an instrument that scientists use to carry out their research quicker and more effectively, rather than something that requires very specific domain knowledge to utilize.

To this end, we created gt4sd-core, an open-source library to accelerate hypothesis generation in the scientific discovery process that eases the adoption of state-of-the-art generative AI. GT4SD includes models that can generate new molecule designs based on properties such as target proteins, target omics profiles, scaffolds distances, binding energies, and additional targets relevant for materials and drug discovery.

The library provides an effective environment for the generation of new hypotheses (inference) and for fine-tuning the models to specific domains using custom data sets (models retraining).

GT4SD's common framework makes models easily accessible to a broader community, like AI/ML practitioners developing new generative models who want to deploy with just a few lines of code. GT4SD provides a centralized environment for scientists and students interested in using generative models in their scientific research, allowing them to access and explore a variety of different models — all of which are pretrained. Consistent commands and interfaces for inference or retraining with customizable parameters harmonize the use across the different models.

The development of problem-specific intelligence is made possible thanks to the automatic workflows enabling retraining with users' own data covering molecular structures and properties. The replacement of manual processes and human bias in the discovery process has important effects on downstream applications that rely on the use of AI models, leading to an acceleration of expert knowledge.

If you want to contribute and get involved check out our guidelines and code of conduct.

We are continuously evolving GT4SD but there is already plenty of stuff you can do with it.

A good starting point is to check out the library and the short introductory guide in its README.md.

You can find all implementation details and some examples in our docs.

Try out the notebooks we prepared to use generative models for various scientific applications.

If you are an end-user and not a developer, just try out some of our pretrained models via a simple web-app (no programming required 👩‍💻)