Skip to content

AI-generated antibiotic blueprint designer revealed

Penn Engineering researchers have crafted and verified a novel 'diffusion model' capable of producing potential antibiotic compounds in a manner analogous to how AI produces images.

Artificial Intelligence Model Crafting Novel Antibiotics
Artificial Intelligence Model Crafting Novel Antibiotics

AI-generated antibiotic blueprint designer revealed

In a groundbreaking development, a team of researchers from the University of Pennsylvania have introduced a new AI tool called AMP-Diffusion. This innovative generative AI is designed to create antimicrobial peptides (AMPs) with bacteria-killing potential.

The tool's design strategy focuses on creating peptides with drug potential, a key aspect that sets it apart. AMPs are short strings of amino acids, the building blocks of proteins. AMP-Diffusion generates candidate AMPs faster and its outputs are more likely to follow the intricate patterns that make peptides effective.

The approach of using a fluent speaker, known as ESM-2, as a coach for AMP-Diffusion streamlines the design process. ESM-2, a widely-used protein language model, already has a rich "mental map" of how real proteins fit together. This strategy allows AMP-Diffusion to leverage ESM-2's rich understanding of protein structure, making the peptide design process more efficient and effective.

AMP-Diffusion uses generative AI algorithms similar to those used for image generation, but refines sequences of amino acids instead of pixels. The denoising process is aided by a coach that keeps it grounded in biological reality, ensuring the outputs remain biologically plausible.

In animal models, the most potent AMPs created by AMP-Diffusion performed as well as FDA-approved drugs, without detectable adverse effects. This is a promising sign for the tool's potential in the fight against antibiotic-resistant bacteria.

The authors of the article about AMP-Diffusion are Jose Juan Almagro Armenteros, Alexander Rosenberg Johansen, Ole Winther, and Henrik Nielsen. Other senior co-authors include Pranam Chatterjee, assistant professor in bioengineering and in computer and information science, and César de la Fuente, Presidential Associate Professor in multiple departments.

By starting with a fluent speaker like ESM-2, AMP-Diffusion can bypass the need for extensive biology teaching. This shortcut allows AMP-Diffusion to focus on designing peptides with a real potential to become drugs.

In conclusion, AMP-Diffusion is a powerful new tool that could revolutionise the production of antibiotics. Its use of generative AI and a fluent speaker as a coach makes it a promising tool in the fight against antibiotic-resistant bacteria.

Read also:

Latest