From the genetic code that underpins diseases to the chemical data driving drug discovery, scientists produce a tremendous number of large datasets, How do we best organize all this data so that we can accelerate scientific discoveries? That is the question that Scripps Research bioinformatician Andrew Su and his lab are tackling.

What is your research focus?

My lab works in bioinformatics, which is at the intersection of biology, computer science and statistics. Modern biology generates data at an unprecedented scale, and bioinformaticians like me see that as an exciting opportunity to leverage that data to reveal new actionable scientific insights.

Roughly half of our work focuses on building bioinformatics tools that accelerate research for both other bioinformaticians and for experimental scientists. Many of these tools revolve around the challenge of organizing biomedical data and knowledge. Our aim is to efficiently and comprehensively capture the current state of biomedical knowledge. That aggregated knowledge in turn feeds computational algorithms based on machine learning and AI to generate new scientific insights.

The other half of the lab focuses on applying the tools of bioinformatics – both tools we develop and ones developed by other computational labs – to generate new testable hypotheses. I love these projects because they are often collaborative efforts with other labs here at Scripps Research. They also touch on a diverse spectrum of science: Alzheimer’s disease, cancer, osteoarthritis, rare diseases, aging, infectious diseases and more. 

Is there a project that you are currently working on that you are particularly excited about?

The bioinformatics field has been enamored with machine learning for quite some time now, but machine learning is dependent on having access to high-quality, well-annotated data. While there are definitely pockets of that high-quality, well-annotated data, in general, the biomedical research community historically hasn’t done a great job of organizing knowledge and data. My lab has put a lot of energy into broadly improving the state of data and knowledge management, an activity sometimes referred to as “knowledge engineering.”  

I’m excited now because the convergence in advances in biomedical knowledge engineering and artificial intelligence is allowing us to make huge strides in knowledge engineering through community-wide collaborative efforts like the NCATS Biomedical Data Translator program, a consortium we’ve been part of since 2016. Combining those efforts with recent advances in large language models (LLMs) is very exciting. Whereas LLMs on their own can be prone to hallucinate, the grounding of LLMs in trustworthy data and knowledge resources is very powerful. I’m very optimistic about an emerging landscape of tools that combine LLM’s general purpose information synthesis capabilities with authoritative biomedical data resources.

What made you want to pursue your research career at Scripps Research?

I actually made the choice to join Scripps Research twice – first as a grad student and then again as a faculty member. They were very different circumstances, but the same basic factor drove my decision. Scripps scientists at every level embrace opportunities for collaborative science. As an incoming graduate student in 1998, I was excited by the interdisciplinary culture. Since I joined the faculty here in 2011, the opportunities for those collaborative projects have only grown. I have been fortunate to work with other Scripps faculty on many research projects and consortia, and those partnerships have been the source of some of our most exciting discoveries and rewarding training experiences.

What does mentoring mean to you?

The challenge of independent scientific discovery is what first drew me to an academic career, but I’ve found the challenge of effective mentorship of trainees to be equally demanding and rewarding. I used to do quite a bit of rock climbing earlier in life, and I see a lot of parallels between being a good rock climbing partner and being a good mentor. First and foremost, I need to be in a position to catch my partner if they fall. I’m also giving encouragement and advice from my vantage point, recognizing that each climber benefits from different feedback in both style and substance. And finally, even if I’m the more experienced person, ultimately it is up to my partner to combine my advice, their own intuition, and what they see in front of them to choose their strategy to climb the rock in front of them. 

To me, all that boils down to a few key principles: acknowledging the value of each trainee’s unique experience and perspectives, working together to build new skills, providing open and clear feedback, and identifying shared goals while also respecting individual career aspirations.