AI-Driven Molecular De-Extinction: A New Frontier in Combating Drug-Resistant Pathogens

Researchers are leveraging artificial intelligence (AI) to mine the DNA of long-extinct species, such as woolly mammoths and giant sloths, to uncover genomic secrets that could help combat today's most infectious pathogens, according to NVIDIA Technical Blog.

Addressing a Growing Crisis

Every year, more than 1.25 million people worldwide die from infections that are resistant to current drugs like antibiotics, as reported by the World Health Organization (WHO). This number is projected to rise to 10 million by 2050. Additionally, within six years, around 24 million people could be pushed into extreme poverty due to the costs associated with treating infectious diseases.

AI and Molecular De-Extinction

Dr. Cesar de la Fuente, a professor at the University of Pennsylvania, is leading a team of researchers to use AI in a process they call “molecular de-extinction.” This technique, detailed in a paper published in Nature Biomedical Engineering in June 2024, aims to identify novel solutions to dangerous drug-resistant microbes by analyzing DNA from extinct species.

“Exploring and comparing molecules throughout evolution can unlock new biological insights,” Dr. de la Fuente explained. “Our AI-driven molecular de-extinction work allows us to bring back molecules from the past to address contemporary challenges.”

Advanced Computational Techniques

Using a cluster of NVIDIA A100 GPUs, Dr. de la Fuente and his team trained deep learning models to mine the proteomes of both living and extinct species. The scientists hypothesized that pathogens, which have adapted to modern-day drugs, might be vulnerable to antimicrobial defenses found in ancient genomes.

The team trained 40 variants of deep learning models, named APEX, on DNA extracted from fossils of extinct animals and plants. These included species such as extinct bears, penguins, and woolly mammoths. The training utilized a combination of 988 in-house created peptides and thousands of publicly available antimicrobial peptides (AMPs) and non-AMPs.

The models, trained using the cuDNN-accelerated PyTorch framework with a single NVIDIA A100 GPU, predicted encrypted peptide sequences—protein fragments that immune systems use to fight infections. APEX predicted over 37,000 peptide sequences with antimicrobial potentials, 11,000 of which were not found in living organisms.

Laboratory Successes

From the APEX-generated peptides, the researchers synthesized 69 potential antibiotics. In lab tests, mice infected with a bacterial pathogen commonly found in human burn victims were treated with these ancient peptides. The results were promising; the experimental antibiotic derived from giant sloths, named mylodonin-2, showed significant improvement in the health of the mice within two days, comparable to those treated with the common antibiotic Polymyxin.

“Exploring extinct organisms allows us to access a vast array of molecules that contemporary pathogens have never encountered,” Dr. de la Fuente said. “Molecular de-extinction can provide a new arsenal of compounds to combat antimicrobial resistance, one of humanity’s greatest threats.”

Future Prospects

The researchers noted that the de-extincted antimicrobial molecules attack microbes by depolarizing the inner membrane of a pathogen’s cells, a mechanism different from most known antimicrobial peptides. This innovative approach, made possible by advancements in AI and GPU technology, seems almost like a plot from a Michael Crichton novel.

Dr. de la Fuente believes that generative AI holds the potential to revolutionize drug discovery methods, reducing both the cost and time required for developing new antibacterial drugs. Traditional methods can take up to 15 years and cost over $1 billion, but AI-driven approaches can significantly shorten these timelines.

“GPUs are transforming how we do our work in our lab,” Dr. de la Fuente said. “We can accomplish in a few hours what used to take six years of research. This has enabled us to dramatically accelerate antibiotic discovery. It’s like bringing science fiction into reality.”

Dr. de la Fuente is in the early stages of setting up a company to commercialize the most promising antimicrobial drugs discovered by his research team. The Machine Biology Group continues to explore promising antimicrobial peptides using their APEX models. Their work is open source and available on GitHub.

For more detailed information, readers can review the Nature paper and other publications from Dr. de la Fuente’s lab.