Q: What made you want to become a scientist?

Dave: My older sister started studying biology when I was 10 years old, and I remember her telling me about the classic phenotype-based screens in worms and flies and I just thought, cool. I loved the idea of these tiny molecules affecting big, observable phenotypes. So, when I reached high school, I volunteered at a research lab at UC San Diego and never really looked back.

Q: When did you first uncover Zebrafish as a compelling model for experimentation and validation of novel drug discovery?

Dave: It all started in grad school in what felt like serendipity. I was doing genetic screens on worms. I discovered that common insecticides, like mothballs, suppressed cell death in worms by inhibiting key enzymes that normally drive cell death. Turns out, the mothballs being used in our building – to keep fruit flies under control – were inadvertently affecting my worms. It was an unexpected discovery that suggested how some insecticides promote cancer. I published the results in Nature Chemical Biology and this influenced my post-grad life in several ways.

First, I became very interested in chemical screens as an unbiased, systematic approach for linking molecules to phenotypes. Second, I began to embrace the role of chance in my research. Third, I went on to pursue a post-doc fellowship at Harvard Medical School to study the intersection of neuroscience and chemical biology. And it was there, while building genetic models of psychiatric disease, I discovered – again, by chance – a novel behavioral phenotype in wild-type Zebrafish that forever changed my thinking about how to approach drug discovery. While other researchers constrained themselves with narrow hypotheses, I developed new screening platforms that were designed to be high-volume, high-scale – and maximize the chances of discovering new drugs.

Q: Super cool. Zebrafish is not a widely used modeling system in drug discovery. Can you tell us more about its recent history and why it’s emerged as an attractive option for biomedical research?

Dave: Zebrafish is a relatively new model system, compared to flies and mice. The first chemical screen of Zebrafish was published in 2000, by Randy Peterson, who is a leader in using Zebrafish in biomedical research and was my post-doc mentor. But the thing that got me excited about the fish was the ability to do chemical screening.

Zebrafish are small and aquatic, which means you can put them into 96-well plates. With mice, the standard of modeling, you can’t test a compound quickly. You’re going to have go much slower. If you’re doing fish screening, you can test thousands of compounds in a day, at orders of magnitude.

Zebrafish are also transparent which allows you to phenotype not just behavior, but also structure of the organism. You can capture data on internal organs, heart, blood, tumors, macrophages. And when you add in modern technologies like robotics, automation, and AI – you can, for the first time, combine large-scale screening with large-scale experimentation. This gives you access to way more systems biology which, hopefully, will unravel more of human disease biology and allow us to better understand and predict disease.

Q: Is that why you joined BioSymetrics? Because of our use of Zebrafish combined with automation and AI?

Dave: That was a major draw. I am thrilled to join a company that understands the power, opportunity, and vision of combining of AI with high-content screening in an intact, vertebrate model system like Zebrafish.

On a personal note, before making the decision to join BioSymetrics, I ran an academic research lab and a biotech startup. At one point, I did both simultaneously! For a while, I thought academic research would give me the freedom to make the biggest impact. But, I found running a startup with academic funding limiting. The key bottleneck in my research was access to engineers and computer scientists. I realized I needed a true technology-first company to support my research. And then I found BioSymetrics.

Q: Well, we’re equally thrilled to have you aboard. Final question: what excites you about BioSymetrics’ approach to drug discovery, or AI-powered drug discovery in general?

Dave: Three things… scale, reproducibility, and iteration. Scale of discovery can compensate for uncertainty. Discovery biology is so complex, we can’t depend on one or a few predictions. We need many to become reasonably confident in our discoveries. Typically, this means a cell-based in vitro assay system. But at BioSymetrics, we’re combining scale with an intact vertebrate (Zebrafish) system and robots. Which is just, awesome.

Reproducibility is key because as a researcher, it is easy to get distracted by noise. Large-scale assays drive reproducible results. In other words, it’s easy to take more shots on goal and that, hopefully, translates to more wins. Scale plus reproducibility help us zero in on rare compounds with strong phenotypes. Ideal.

Finally, iteration allows us to learn and improve. Machine learning helps close the prediction-experiment loop and feed us data, quickly. I love the approach to feeding an algorithm with high-quality, high-content data and using computational power to investigate predictions that I could have missed by eye. AI-assisted research is exciting and liberating because it offloads certain aspects of pattern-finding to the algorithm. This really frees up one’s brain to get more creative about problem solving and moving forward.