Searching for new antimicrobial molecules with code and AI

We haven't discovered a fresh class of antibiotics in half a century. Think about that for a second. While drug-resistant pathogens quietly evolve into an existential crisis that claims millions of lives annually, our traditional R&D pipeline remains stuck modifying the same tired chemical families over and over again. It is a slow, expensive loop of diminishing returns. The entire medical establishment has been digging in the exact same dirt for decades, expecting a different result.

Fortunately, labs like César de la Fuente's are flipping the script by treating biology not as a mystery box of physical matter — but as code. This dNA nucleotides and amino acids form a massive, largely unread alphabet. And you can stop blindly splashing chemicals onto petri dishes. And start parsing the underlying logic of life itself when you view genomes as info systems. Real talk: deep learning models can scan vast biological sequence databases to hunt for signals that human eyes would miss in a lifetime.

What fascinates me isn't just the custom neural nets trained on specific protein datasets, but how everyday developer tooling accelerates the breakthrough. Researchers are leaning on models like ChatGPT and Codex to brainstorm biological hypotheses, write pipeline scripts, parse messy datasets, and bridge disparate scientific domains on the fly. It is software engineering applied to wetware. Code is becoming the ultimate laboratory assistant.

This is what actual, high-impact computing looks like. Too much of our industry is obsessed with building wrapper apps for writing marketing copy or generating mediocre art, while genuine problem solvers are using these exact same cognitive engines to rewrite the rules of pharmacology. They are compressing centuries of tedious trial-and-error chemistry into a matter of hours.

We desperately need more of this energy. Technology matters most when it tackles hard, physical problems that genuinely threaten our collective future. The tools are finally sharp enough. Now we just have to point them at the right targets.