Available Resources

Machine learning frameworks

Our machine learning (ML) frameworks use data from biophysical codes and other sources to conduct cutting-edge experiments.

available-resources-workflow-focused-codes

Workflow development

CPB-designed tools allow us to create highly optimized workflows in the Livermore Computing (LC) environment. These workflows are used to generate data as well as visualize and analyze the data.

Data science

Our machine learning frameworks and unique data utilization tools let us generate unique hypotheses for a very broad range of challenges in material sciences, healthcare, and biosecurity.

simulations-and-analysis-tools-infographic

Simulation and analysis

We use HPC enabled Molecular Dynamics engines and docking tools that run at scale in the LC environment. Visualization and analysis tools are available for extremely large datasets.

Data ecosystem

LLNL is building a robust data ecosystem to enable development and distribution of high quality, hard-to-reproduce data for external partners in academia, industry, and other DOE laboratories. LLNL currently has data ecosystems in place for local data sharing, but the long-term plan is to extend them into an ecosystem that enables internal and external collaboration on data.

Research areas supported

Human genomics
Pathogen genomics
Chemical structure and descriptors
Protein structure and interactions
Knowledge networks

Data types

Machine learning modeling
Experimental
Physics-based simulation
DNA sequence databases
Protein sequence, structure, and interaction databases
Chemical-biology interaction databases

Data portals

Experimental platforms

We are developing an ultra-high-throughput data generation engine to produce high-quality datasets at unprecedented scale to enable training of generalizable machine learning algorithms for biologics design.

With an initial focus on high-resolution profiling of the antigen/antibody interaction landscape, our ongoing methodological development and integration with active learning will enable characterization and rapid AI-based design of powerful biologics.