Caltech has received a $17 million grant from the National Science Foundation to build ELECTRA, an automated laboratory that will use artificial intelligence to determine the three-dimensional structures of millions of molecules remotely. The facility represents a fundamental shift in how chemists approach one of their field's biggest unsolved challenges: characterizing the vast universe of molecular structures that remain unknown.

What Is the "Chemical Dark Matter" Problem?

Millions of molecules exist in nature and in chemical libraries, but their precise three-dimensional structures remain a mystery. Caltech professor Hosea Nelson, who leads the ELECTRA project, compares this gap to dark matter in physics, noting that understanding these structures is essential for unlocking new drugs, advancing biology, and discovering novel materials. The challenge has persisted because traditional methods for determining molecular structure are slow, expensive, and require large quantities of sample material.

Nelson explained the scope of the problem: "We have tools that allow us to make guesses at many structures, but this project is about revealing all of those 3D structures on a large scale." He frames ELECTRA as analogous to the Human Genome Project, but for natural products and chemical compounds. Historically, determining a single molecular structure was groundbreaking work; in the 1920s, such discoveries earned Nobel Prizes. Today, the goal is to accomplish the same task in under a minute.

Nelson

How Will ELECTRA Use AI and Automation to Speed Up Discovery?

The lab will employ microcrystal electron diffraction (microED), a technique that Nelson's group pioneered in chemistry by adapting methods from structural biology. MicroED excels at resolving atomic arrangements even with tiny sample sizes, a critical advantage when working with rare natural products or compounds available in limited quantities. The real innovation lies in combining this technique with artificial intelligence and robotic automation to process samples at scale.

ELECTRA will operate as a cloud-based facility, meaning researchers across the country can submit samples remotely and access the resulting structural data. The lab will house robotic systems for high-throughput experimentation, allowing it to process far more molecules than traditional chemistry labs. The AI expertise of Caltech co-investigators Katie Bouman and Yisong Yue will automate data analysis and storage, turning raw experimental results into usable structural information.

The collected data will serve as a training ground for increasingly sophisticated AI models. Unlike protein structure prediction, where large datasets already exist, chemistry has lacked comparable resources for training algorithms. ELECTRA will help fill that gap, enabling AI systems to predict molecular structures from limited data and further accelerate discovery.

Steps to Leverage ELECTRA for Your Research

• Submit Samples Remotely: Researchers nationwide will be able to send molecular samples to ELECTRA through its cloud-based platform, eliminating the need for on-site access to expensive equipment.
• Access Structural Data: Once ELECTRA determines a molecule's three-dimensional structure, the results will be available through a shared data repository that researchers can query for their own work.
• Train AI Models: Scientists can use the growing database of solved structures to develop and refine machine learning models that predict molecular properties and behaviors.

What Real-World Applications Could ELECTRA Enable?

The implications extend across multiple scientific disciplines. In biomedicine, understanding the structures of natural product molecules could unlock new pharmaceuticals. Taxol, a widely used chemotherapy drug, is derived from the Pacific yew tree; acetylsalicylic acid, the active ingredient in aspirin, comes from willow bark. Many other natural compounds with therapeutic potential remain structurally uncharacterized, waiting for tools like ELECTRA to reveal their secrets.

Beyond medicine, ELECTRA's scope encompasses materials science, geology, and fundamental chemistry research. The ability to rapidly determine molecular structures could accelerate the discovery of novel materials with specific properties, from stronger alloys to more efficient semiconductors. The lab will also support investigations into complex molecular interactions within biological systems, potentially unlocking new insights into disease mechanisms and therapeutic targets.

"ELECTRA represents a bold move in accelerating scientific discovery," said Caltech President Ray Jayawardhana, envisioning a resource that will fuel breakthroughs across multiple scientific fields.

Ray Jayawardhana, President at Caltech

Who Is Behind This Ambitious Project?

ELECTRA is a collaborative effort bringing together experts from diverse institutions. Beyond Caltech's Hosea Nelson, Katie Bouman, and Yisong Yue, the team includes Jose A. Rodriguez of UCLA, Garret Miyake of Colorado State University, Emily Balskus of Harvard University, and Alison Narayan of the University of Michigan. Nelson emphasized the importance of this interdisciplinary approach: "We don't know what these molecules are in many cases, so we pulled together experts in disparate fields to help us build out a platform that can apply to all of them".

Nelson, Katie Bouman, and Yisong Yue, the team includes Jose A

The $17 million National Science Foundation grant reflects confidence in the project's potential to transform how chemists approach molecular discovery. The facility is expected to dramatically reduce the time required to determine molecular structures, turning a process that once took decades into one that takes seconds. This acceleration could unlock a vast reservoir of chemical knowledge and accelerate innovation across materials science, drug discovery, and fundamental research.