The Moonshot Podcast S2, Episode 2: Coding The Natural World
This episode of The Moonshot Podcast delves into the future of biological engineering, showcasing how AI and computational biology are transforming our interaction with living systems. Host Astro Teller first speaks with Brad Zamft of Heritable Agriculture about programming plants for increased yield, pest resistance, and drought resilience. Next, Relly Brandman from project A-Life explains how they're using AI to create a "virtual cell," shifting biomanufacturing from slow trial-and-error to a predictable engineering discipline for producing diverse materials like medicines, fuels, and textiles.
Introduction: Biology as an Engineering Discipline
Host Astro Teller introduces the idea of biological systems as “tiny factories” that evolution has been perfecting for billions of years. He highlights two examples: a seed that becomes a plant, and a single cell. The central question posed is whether humanity can edit these biological units to unlock new opportunities. He states that while traditional industrial processes make trillions of dollars worth of goods, the future could see biology used to create medicines, chemicals, and fuels, moving biology from a science to an engineering discipline.
Heritable Agriculture: Programming Plants for a Sustainable Future
The Agricultural Challenge
Brad Zamft, CEO of Heritable Agriculture, discusses the immense success of the $3 trillion agricultural industry in feeding a growing population with increasing quality of life. However, he emphasizes that past methods of improving crops, primarily traditional breeding, “won’t get us there” in the face of future population growth and climate challenges. A “step change” in optimizing the agricultural system is needed.
Reprogramming Plants: Beyond Traditional Breeding
For 9,000 years, humanity has been “programming plants” through selective breeding, biasing the “random walk in the genome space.” More recently, with DNA sequencing, scientists can build models to predict plant phenotypes (physical properties) from genotypes. Heritable Agriculture’s innovation goes beyond merely reading DNA, which is static and inherited. Instead, they focus on other molecular information like RNA, metabolites, epigenetics, and proteins, which “are moving in real time, responding in real time to the environment.” This provides a much richer, dynamic understanding of how a plant is experiencing the world.
Early Innovations and Challenges
Heritable Agriculture’s journey began at X, where they started with model organisms like Arabidopsis thaliana, a fast-growing plant with no commercial value but well-characterized biology. They aimed to change a simple trait like leaf number, extracting DNA and RNA to build models. These models successfully identified genes associated with leaf number, and subsequent gene editing validated their predictions. They then applied this approach to corn, discovering new flowering time genes in just 18 months. Experiments were conducted in sophisticated growth chambers at X, and even on the building’s roof. Scaling to real-world conditions presented challenges, such as conducting time-sensitive RNA sampling of corn roots in the sweltering Wisconsin heat during the pandemic.
From Moonshot to Independent Company
Heritable Agriculture transitioned from an X project to an independent company. Brad highlights that pitching many investors instilled discipline and rigor. He emphasizes the importance of culture, viewing it not as a 9-to-5 job but as a passion, fostering an environment where people feel valued and cherish.
A Vision for Future Agriculture
Brad’s 10-year vision for Heritable Agriculture is to see an improved plant in the ground that enhances the nutrition of the global food system. He hopes to make a significant “dent in the efficiency of the agricultural system” and improve the lives of consumers and growers. He envisions a future where the sophistication currently applied to corn and soy for yield optimization is extended to all fruits, vegetables, forests, and other plant-based materials, enabling them to provide food, nutrition, fiber, fuel, and durable materials like wood.
Project A-Life: Unlocking Biomanufacturing with AI
Biology as Tiny Factories
Relly Brandman, project lead for A-Life, describes A-Life’s mission to “unlock the manufacturing power of biology.” She explains that single-celled organisms like yeast naturally produce substances (e.g., alcohol in beer/wine), and with advanced genetic editing tools developed over recent decades, these cells can be reprogrammed to create a vast array of materials: food, polymers, fuels, textiles, building materials, vitamins, and medicines. Examples include enzymes in laundry detergents that enable cold-water washing. Cells function by taking in sugar and nutrients, converting them into energy, and performing biosynthetic pathways to grow or produce desired materials, often expelling them as what the cell perceives as “garbage.”
The Need for Predictive Design
Despite the potential, biomanufacturing is currently slow and expensive because scientists don’t predictably know the outcomes of genetic changes. The process involves extensive trial-and-error at lab scale, followed by another round of optimization for large-scale fermentation in swimming pool-sized tanks. This leads to costly late-stage failures. A-Life aims to shift this process from trial-and-error to “predictive design,” treating the entire system from bench to manufacturing as a single optimization problem. Relly notes that many complex molecules, especially protein-based medicines and antibody drugs for cancer therapies, can only be made using biological systems, making this advancement crucial.
The A-Life Approach: Building a “Virtual Cell”
A-Life addresses the unique challenges of biological data: it’s typically small and highly heterogeneous, unlike the vast, uniform datasets common in internet applications. Their approach involves encoding known cellular biology into a computer program. In their lab, scientists edit DNA, grow cells under various conditions (from tiny plates to 2-liter fermentation tanks), and use analytical chemistry equipment (like HPLC and mass spectrometers) to take “cell signature snapshots.” This involves breaking down biological samples to identify and quantify millions of components (proteins, lipids, metabolites), providing the necessary data for AI models to learn.
AI-Guided “Design-Build-Test-Learn” Cycles
A-Life creates a “virtual cell” that models how a cell functions within its environment, considering both internal genes and external conditions (nutrients, pressure, temperature). This virtual cell allows them to run millions of simulated experiments, changing genes or environmental factors, to predict outcomes. This predictive capability guides the lab’s experimental scientists, prioritizing the most impactful experiments in what they call “design-build-test-learn” cycles. AI guides the design (simulating millions of options) and the learning (updating models with new data), significantly reducing the number of physical experiments needed to achieve better results.
Impact and Future Vision
For customers—expert biomanufacturers—A-Life integrates AI into their workflows, simulating experiments, prioritizing next steps, and making recommendations. The goal is to move from manual pipetting to robotic systems to scale up data collection significantly. Relly believes that the “virtual cell” could evolve beyond material engineering, serving as a platform for understanding life itself, such as microbiomes and human health. The challenges of building A-Life were met with support from X, including custom tools from the Design Kitchen for rapid, precise sampling and freezing of biological material. Relly’s 20-year vision for A-Life includes local biomanufacturing factories with lower carbon footprints, producing entirely new materials, including “living medicines” and “self-healing bricks.”
Conclusion: Harnessing Technology for Nature
Astro Teller concludes by reiterating the profound shift of biology from a science to an engineering discipline. He emphasizes that by using technology to help nature, humanity can more effectively address rapidly emerging global challenges, underscoring his excitement for both Heritable Agriculture and A-Life’s contributions.