Research areas
Early warning and threat assessment
The Center for Predictive Bioresilience seeks to rapidly detect, identify, and characterize unknown threats, whether natural or human-made, by expanding the limits of genomic sequencing. We will overcome these limitations through a combination of computing, software, data, and experimental systems.
Viral evolution is a major challenge to protecting against infectious disease. Viruses are capable of rapid mutation in response to a variety of environmental changes, such as adapting to new hosts, evading immune response, and escaping therapeutic treatment.
Viral forecasting predicts the mutants more likely to emerge from circulating strains. Previous work at LLNL improved methods to detect new mutations by using deep sequencing and by mapping mutations onto protein structure to assess their impact on function.
Current projects are modeling the impact of mutations on viral fitness and adaptation using a combination of experimental data and different computational methods, bringing us closer to effective, reliable mutation forecasting and protection against infectious disease.
The goal for detector research at LLNL is to develop cost-effective early warning systems of emerging biological threats. Lawrence Livermore Microbial Detection Array (LLMDA) supports biosecurity with the capability to detect 12,000 known microbes, including viruses, bacteria, and fungi.
Work at CPB is using statistical models and bioinformatics methods to identify high-priority pathogens, so rapid detectors can be developed and deployed for routine biosurveillance applications.
LLNL has also developed field-deployable detectors with a smaller range of pathogen identification. Bio ID uses isothermal technology to rapidly identify up to 20 biothreat agents, displaying results in a compact, portable format that makes it ideal for point-of-care, laboratory, field, and low-resource settings.
The detection of uncharacterized, emerging, and over-the-horizon biothreats presents a significant challenge for existing biodetection and biosurveillance systems. LLNL research addresses this critical gap by integrating sample processing techniques, bioinformatics tools, and predictive models to rapidly identify and characterize threats across diverse sample types.
CPB is particularly focused on improving the sensitivity of genomic and proteomic sequencing to discover previously undetected sequences and classify them into pathogenic threat families. In instances where sequences remain unclassified, our computational pipeline aims to identify their disease-causing potential at the peptide and protein levels.
This analytical process allows researchers to predict whether uncharacterized sequences are anomalous and may pose a human health threat.
Metagenomics analyzes and characterizes DNA sequences to identify unknown microbes and understand their functions. LLNL researchers use local databases containing nearly every known DNA genome to analyze multiple genomes simultaneously, the scale of which requires overcoming major challenges in data and computing.
Using metagenomics, we can identify unknown infections and improve treatment management for infections and wounds. The goal is to give more information to healthcare providers and policy makers to guide health decisions and maximize successful outcomes.
Nanolipoprotein particles (NLP), also known as nanodiscs, are nanoscale membrane mimetics consisting of lipids (green) stabilized by apolipoproteins (blue).
BBO-8520 (a first-in-class compound in clinical trials) and GTP bound in KRAS G12C.
Accelerated medical countermeasure design
Designing safe and effective countermeasures in just weeks instead of years allows us to bring lifesaving drugs to market more quickly and optimize their ability to address health needs. We use computing, automation, and additive manufacturing to radically shorten the drug design-make-test cycle and optimize therapeutics.
Rapid antibody design with GUIDE
Countermeasure design begins with protein and antibody design, an area in which LLNL has tremendous capability via the Generative Unconstrained Intelligent Drug Engineering (GUIDE) program. GUIDE was initiated during antibody development for the Omicron variant of SARS-Cov-2, and it is now primed to address future biological threats.
Small molecule drugs, the most common type of therapeutic, are lifesaving for many patients. However, drug development is expensive and time-consuming. Livermore has integrated machine learning and physics-based modeling and simulation to create a design and optimization platform for drug development that accelerates the development timeline.
In addition, we have created a new platform to explore unknown protein targets and identify cryptic binding sites, which require significant rearrangement of the protein structure to become physically accessible to a targeting agent. The goal is to further expedite new drug discovery and development.
We collaborate with other federal agencies, biotechnology companies, and pharmaceutical partners to develop our platforms. This has led to the co-design of a first-in-class drug candidate with an approved IND filing for dosing in humans.
Host-targeted antivirals interfere with the host genetic factors that viruses leverage to establish or propagate an infection. LLNL is contributing to host-targeted antiviral development efforts primarily through the PROTECT initiative.
PROTECT is a computational and experimental collaboration between LLNL, UCSF, the University of Texas Medical Branch, and the Innovative Genomics Institute of UC Berkeley to develop host-targeted therapies for human viral infections using reversible epigenetic editing techniques. Its goal is to develop a short-acting antiviral with a reversible epigenetic silencing effect that disables an essential host factor used by multiple related viruses.
Experimental human models are physical experimental models with multiple data read-outs that mimic human organs and organ systems like the brain, lungs, and gut. Livermore is developing and using these models to understand the human body’s reaction to mission-relevant exposures, such as opioids, pesticides, or pathogens like SARS-CoV-2.
Developed in collaboration with biologists, engineers, and computational scientists, human experimental platforms are used to safely test hypotheses and understand mechanisms of exposure in advance of such threats.
LLNL is developing a platform to enhance immune response and improve efficacy for a wide range of vaccines. Through this platform, the components of a vaccine are assembled on a nanoparticle, ensuring that multiple copies of each critical element are presented. This creates a more effective, evenly-dispersed solution than is made by simply mixing constituents together.
Using this approach, the immune system is simultaneously exposed to the pathogen-specific antigen and the immunostimulatory adjuvant molecules, made possible by the scalable and functionalizable nature of our nanoparticles.
The LLNL vaccine platform can accommodate antigens from any pathogen in a “plug-and-play” approach, providing the flexibility and speed needed to develop vaccines against new and emerging pathogens. In addition to vaccines for human health and biosecurity applications, our nanoparticles can be used across multiple applications in biology and health sciences, including the delivery of drugs and therapeutics.
Safety and security research and development for AI-guided biosystem design
Biosystems, such as cells and biological pathways, can be designed for use in specific applications with machine learning. The Center for Predictive Bioresilience will promote research and development around the safety and security of AI-guided biosystem design. We will use AI to address existential threats while mitigating new risks generated from deliberately misused or unsupervised AI.
A pilot program will bring “fellows” into the Center to conduct research on integrated AI-biology safety and security from both technical and policy perspectives. The Center for Predictive Bioresilience will set the standard for responsible conduct in AI-guided biosystems design. Specific R&D includes:
- Risk assessment tools and frameworks
- Confidence limits and uncertainty quantification in predictive models
- Data provenance and use assessment
- Attack surface characterization for biodesign systems
- Risk modeling and policy development research
Work with the Center for Predictive Bioresilience
Collaboration is a key element of CPB’s success. Learn more about how we work with industry, academic, and national laboratories to extend and refine our unique approach to medical countermeasure design.
