Â鶹ÒùÔº

Imagine a bacterial cell—one of the that keep infectious disease experts up at night—blown apart like a microscopic firecracker.

That's exactly what scientists are seeing when a new kind of molecule called SNAPP meets bacteria.

It's not science fiction—it's molecular dynamics in action.

Our team is using molecular dynamics computer simulations to understand how —Structurally Nanoengineered Antimicrobial Peptide Polymers—destroy drug-resistant bacteria.

We're hoping to change the game in the battle against the escalating .

A jellyfish that kills bacteria

SNAPPs resemble tiny jellyfish—soft, spherical bodies with wavy amino acid 'arms' that reach out and interact with the membranes that surround bacterial cells.

And those arms are deadly.

Even at low concentrations, SNAPPs induce rupture, collapse and disintegration of bacterial cell membranes. Rather than simply attacking the bacteria, SNAPPs destabilize the fundamental structural integrity of the membrane.

This eliminates any chance of the bacteria developing resistance against SNAPPS.

Our collaborators, , first demonstrating their antimicrobial activity in the laboratory.

But not all SNAPPs behave the same way. Small changes in the sequence, type and number of their amino acid arms can change how effectively they kill bacteria—and whether they might also harm human cells.

And there are lots of possible SNAPPs.

There are 20 types of amino acids to choose from and there are hundreds spread across the eight arms of each SNAPP molecule. That means countless possible combinations.

It's a molecular recipe with more variations than any kitchen cookbook.

Building a smarter way to design molecules

That's where our team comes in.

At the group, we combine computational materials science with machine learning and eXplainable Artificial Intelligence () to tackle complex design challenges like SNAPPs.

We've built that let us see what no experiment can: the moment-by-moment interaction between a SNAPP molecule and a bacterial cell membrane at the nanoscale (billionths of a meter).

Ph.D. student Amal Jayawardena has spent the past couple of years that reveal how SNAPP arms sink into bacterial membranes, fold around them and essentially tear them apart from the inside.

It's stunning to watch and it illustrates why even a few SNAPP molecules can be lethal to bacteria.

Just as important, we can see how these molecules don't latch onto mammalian cells the same way. That's because bacterial membranes carry a distinct surface charge—a feature our cells lack.

That's part of what makes some of SNAPPs so promising as targeted antimicrobials.

Enter Bandicoot: AI meets molecular design

Of course, simulating one SNAPP at a time won't get us far.

There are too many possible combinations. We need a system that knows where to look and what to look for.

So, we're building it.

We call it Bandicoot. We've named it after that poke around in the soil, sniffing out what's hidden beneath the surface.

That's what Bandicoot does: it explores complex data, identifies where our knowledge is thin and directs simulations and experiments to the most promising areas.

Postdoctoral researcher Dr. Nick Robe built . We've adapted it to create thousands of different SNAPP variants.

And Ph.D. student Jalal Abdolahi has developed —tools that help Bandicoot decide which SNAPPs are worth simulating next.

If some variations aren't changing outcomes, we don't waste time on them. Instead, we focus where the science is hottest.

More than a 'black box'

But we don't just want Bandicoot to tell us what works. We want it to tell us why.

That's where comes in.

Instead of a black box that spits out answers, Bandicoot gives us insights—explaining which parts of a SNAPP molecule make it effective and how those features interact.

That kind of interpretability is critical. It helps scientists like us understand the fundamental rules behind molecular behavior.

It also has the potential to help product developers create safer, more reliable treatments.

These explanations don't just satisfy curiosity—they also help speed up the discovery process.

Turning the tide

Antibiotic resistance is a global health crisis. But with tools like SNAPPs and Bandicoot, we're shifting from reaction to prediction—from fighting fires to designing fireproof systems.

For the first time in decades, the upper hand may be within reach.

Smart molecules made by smarter systems, with the potential to outsmart superbugs at their own game.