Mining millions of genomes for the next powerful antibiotic

Erik Wright, assistant professor of biomedical informatics at the University of Pittsburgh, spoke to Fearless Science Magazine about how his quest to discover new antibiotics to counter resistance — and how that pursuit has made him biology’s No. 1 user of the Open Science Pool for advanced computing.

How did an electrical engineer end up as a microbiologist?

After a few years working at Apple as an electrical engineer, I decided to go back to grad school and switched to environmental engineering. Studying water and wastewater treatment introduced me to microbes and bacteria and I just fell in love with it. I shifted to microbiology for my Ph.D. at UW–Madison, and the prerequisite classes felt like I was being fed gold nuggets of information. There’s this whole invisible universe out there of microorganisms that I was completely oblivious to. I found a little niche for myself doing biology tied to computing. And now I run what I call soggy lab, which is a hybrid wet and dry lab together.

Why did you decide to make antibiotic resistance the focus of your research work?

The bugs evolve resistance to our drugs. It’s a really hard thing to counter, and it’s especially hard to reverse. I like that it’s so difficult, that’s probably the main draw. I live and breathe the idea that resistance is something that can be stopped and reversed. I began to study the natural antibiotic producers, the microorganisms that we get about 70% of our antibiotics from. These organisms have been naturally producing antibiotics for about half a billion years at least, and because they’re mostly bacteria, they’ve also figured out how to resist them.

What are the areas of focus for your lab?

We’re one of the few labs that studies what strategies bacteria use to avoid resistance. Then we want to understand how to use to avoid resistance. Then we want to understand how to bring that strategy to the clinic and scale it up. We are studying durability, to understand why some antibiotics have been able to avoid resistance. And we are exploring how we can prescribe them in a multidrug cocktail such as for HIV and tuberculosis. The vast majority of the antibiotics we give are pure compounds in a high dose, but that’s only one hurdle for the bug to jump over, we want to present them with many hurdles. We’re trying to figure out how to work with the existing available pool of drugs to do something that’s better than what we currently do. And we think that by changing the way we treat patients — mimicking the biology that currently exists — then maybe we can figure out a more sustainable solution.

You are the number one biology user of high throughput computing with the Open Science Pool. How do you integrate computational approaches to tackle antibiotic resistance?

I had the extreme advantage of being part of the Wisconsin Institute for Discovery at UW–Madison, so I was an early adopter and that has completely changed my career. I’ve been using the Open Science Pool for 12 years and we simply could not reach the kind of computing capacity we need without it. Because it’s open, we don’t have to write grant proposals, which allows us to do a lot of exploratory work. I can’t overstate how much that is worth to me. It is also set up in a way amenable to my research. Instead of shared memory computing, the OS Pool is set up with a distributed memory. We like to split up our work into tiny compartments that each last an hour, so we hit something like 17 million jobs last year.

Why is your lab so computationally intensive?

The main thing my lab’s doing is comparing genomes by processing huge data sets in millions of separate computing jobs on the grid. We have access to about 2 million bacterial genomes, and we have developed software that can draw on thousands of computers to quickly compare new genomes to those that already exist. Then other computers store the data, and thousands more process and analyze the results — all through this gigantic set of grid jobs that is always running. We end up having groups of genes that are the same gene across different organisms, and then we build software that tells us which genes work together. From that we can do things like find which groups make natural products like antibiotics. We aim to build an ecosystem that will live on the grid, which is a little bit of a wild idea, but it will continuously update and compute new genomes and add them into a giant comparison of all genomes versus all genomes. What we’re ultimately going to do is take those groups of genes that work together, transplant them into a host organism, and then turn them on and see what product they make.

What would a dream outcome look like from the research that you’re doing now?

I would like to discover new small molecules that no one has known about. And to find totally new drugs if I could. We’ve developed ways of finding genes that work together, that nobody’s seen before. But we have no idea what antibiotic or other compound that makes. It’s on the order of millions of different possible compounds and it’s a dream to bring some of those to reality. We have developed ways of handling hundreds of thousands of genomes, approaching millions, so this is very much possible.

This feature was originally published by the Morgridge Institute for Research in their “Fearless Science” magazine, Fall 2024.