Lecture will be held in English

Hosted by

Robert Vácha

About the lecture

Mining the Code of Life for New Antibiotics

Biology is the most powerful technology on Earth, yet we still study it largely by hand. That is beginning to change. We now have access to planetary-scale biological data, increasingly standardized measurements, and the computational power to learn from them. AI allows us to compress this complexity into usable representations: models that do not simply describe life, but help us predict, design, and engineer it.

In this talk, I will argue that biology is becoming a digital science, one in which learning algorithms can turn sequences into discoveries at a pace commensurate with the urgency of humanity’s greatest challenges. Antibiotics offer a compelling proving ground. For more than a century, antibiotic discovery has relied on “dirt mining”: slow, local sampling coupled with brute-force screening. That paradigm cannot keep pace with the global rise of antimicrobial resistance. Our lab has been developing an alternative: digital discovery, in which the tree of life becomes a searchable and programmable space for therapeutic innovation.

To make biology programmable, we need a unit that is both information-rich and experimentally scalable. Peptides, the smallest functional biomolecules in life, provide an ideal foundation for this transition. They are tractable to synthesize and test, amenable to rapid iteration, and defined by an astronomical sequence space that remains largely unexplored. I will describe our early efforts using evolutionary computation to design new antibiotics; how we helped pioneer AI-driven antibiotic discovery, generating molecules with strong efficacy in preclinical animal models; and how we then used algorithms to systematically mine the human proteome, revealing thousands of antimicrobial peptides hidden within proteins. These molecules appear to constitute a previously unrecognized layer of host defense that we termed encrypted immunity.

From there, we extended this framework across evolutionary time. By mining ancient biology, we discovered therapeutic molecules encoded in extinct organisms, including Neanderthals and the woolly mammoth. This work helped launch the field of molecular de-extinction and yielded antibiotic candidates such as neanderthalin, mammuthusin, and elephasin.

Next, we expanded across the full breadth of life. By analyzing global microbiomes at scale, we identified nearly one million candidate antibiotic molecules and released them as open access resources to accelerate worldwide collaboration. I will highlight discoveries from human microbiomes, including the molecule prevotellin-2 from Prevotella copri, as well as our most recent step: the digital mining of Archaea, an underexplored domain of life, leading to the discovery of a new class of antibiotics that we call archaeasins. Together, these efforts have dramatically accelerated antibiotic discovery, reducing the time required to identify preclinical candidates from years to just a few hours.

Finally, I will introduce our latest AI stack—APEX, ApexGO, ApexDuo, and ApexOracle—which enables sequence-to-function prediction, computational optimization, multimodal therapeutic design, and rapid-response discovery against emerging pathogens. Across these efforts, we have paired computational discovery with experimental validation, showing that AI can not only surface candidates at scale, but also deliver therapeutically relevant molecules. I believe we are entering a new era in which AI can help us stay ahead of antibiotic resistance, respond more rapidly to outbreaks and future pandemics, and fundamentally accelerate discovery across biology and medicine. 

Registration for lunch with the speaker /for Ph.D. students/

The sponsored lunch usually takes place in the Campus River restaurant. Please meet the speaker and other students at 12:45 at the reception desk at the main entrance (building B22, see the map below).