Skip to content

Unlocking the Power of Open Accelerated Discovery

In the rapidly evolving landscape of scientific research, the need for accelerated discovery has never been more pressing.

With the increasing complexity of molecular systems and the growing importance of data-driven decision-making, researchers are seeking innovative ways to streamline their workflows and unlock new insights.

Enter Open Accelerated Discovery (OpenAD), a revolutionary open-source framework that integrates cutting-edge tools, models, and applications into a single, user-friendly platform. By leveraging the power of AI, machine learning, and advanced computing, OpenAD empowers scientists to accelerate their discovery processes, reduce costs, and enhance productivity.

What is Open Accelerated Discovery?

OpenAD is an open-source project that democratizes IBM's research in machine learning and advanced compute for molecular discovery. This framework provides a low-code offering that delivers accelerators to your molecular discovery, making it easy to deploy scientific AI models and integrate them into your workflow.

With OpenAD, researchers can combine resources from IBM, third-party projects, open-source libraries, or their own custom toolkits into a single pipeline. This flexibility enables scientists to orchestrate their workflows from start to finish, cataloging user-defined or open-source model services, launching and managing model services, and lowering costs of managing services.

Key Features of OpenAD

  • Model Service: Easily deploy scientific AI models without custom-engineered environments.
  • Visualization Tools: Visualize molecules and their properties, view and subset molecules to work with.
  • Built-in Help: In-application help with automated population for chosen model services, including LLM-assisted help powered by IBM Granite LLM on Ollama.
  • Collaboration: Exchange data with other researchers. Easily export and import data and share with others.
  • Data Management: Organize data in project workspaces.

Property Generation Capabilities from Base Models

OpenAD provides a wide range of property generation capabilities from base models, including:

  • Molecules: molecular weight, number of aromatic rings, number of H-acceptors, etc.
  • Proteins: charge density, protein weight, isoelectric point, etc.
  • Crystals: absolute energy, bulk moduli, Poisson ratio, etc.
  • Data Set Generators/Transformer Generation Models
OpenAD includes a variety of data set generators and transformer generation models, such as:

  • PaccMannVAEGenerator
  • MoLeRDefaultGenerator
  • TorchDrugGraphAF
  • TorchDrugGCPN
  • Conditional Generation

OpenAD also supports conditional generation, with models like:

  • AaeGenerator
  • PaccMannRLProteinBasedGenerator
  • RegressionTransformerProteins
  • RegressionTransformerMolecule
  • Controlled Sampling


Finally, OpenAD includes controlled sampling capabilities, such as the CatalystGenerator.

Advantages of OpenAD

  • Easy Deployment: Automate access and deployment of scientific models on the cloud without custom-engineered environments.
  • Faster Iteration: Accelerate machine learning pipeline development.
  • Access to Research Assets: Access scientific research assets through Deep Search, PubChem data, scientific patents, and research papers.
  • Orchestrating Tools and Projects

OpenAD enables researchers to create and manage project workspaces, manage molecular data sets, enrich with model or external data, and integrate with existing tools and processes (e.g., Jupyter notebooks).

Conclusion

Open Accelerated Discovery is a game-changing framework that empowers scientists to accelerate their discovery processes, reduce costs, and enhance productivity. By leveraging the power of AI, machine learning, and advanced computing, OpenAD provides a low-code offering that delivers accelerators to molecular discovery, making it easy to deploy scientific AI models and integrate them into your workflow.

Whether you're a researcher looking to streamline your workflows or an entrepreneur seeking innovative solutions, OpenAD is an open-source project that democratizes IBM's research in machine learning and advanced compute for molecular discovery. Join the community today and unlock the power of accelerated discovery!  For more information visit https://accelerate.science/projects/openad.