At first, Yue Qin thought she wanted to become a doctor. She’d always been interested in disease — why people got sick, why some illnesses could send you to a hospital while others could be treated at home. As she grew older, however, she realized she was more interested in learning about the roots of disease and the genes that caused them. 

Growing up in Ningbo, a city in eastern China, Qin was strongly influenced by societal expectations that girls were better suited for language arts than math and science. She assumed that her struggles in a high school computer science course were because of her gender, even though she’d had no trouble with math. That all changed in college, at the University of California, San Diego (UCSD), when Qin took an introductory computer science course and fell in love with coding and its logic. She began seriously considering a career in computer science and combining it with her interest in biology.

Doing computational biology research as an undergraduate, as well as having supportive research advisors, inspired her to stay in research. After graduating with a bachelor’s degree in bioinformatics, Qin then went on to pursue her PhD at UCSD, where she used computational modeling to study how proteins interact with each other and assemble into a human cell. In January 2023, she joined the Eric and Wendy Schmidt Center at the Broad Institute of MIT and Harvard as a postdoctoral fellow in the labs of Paul Blainey and Caroline Uhler. At the Broad, she aims to create an in silico cell: a computational model scientists can use to study at scale how external influences such as drug treatments affect cells.

We spoke with Qin about finding her place in science, how computational tools can advance biological research, and what it’s like to do both computational and wet lab experiments in this #WhyIScience Q&A. 

What do you like about the intersection of computer science and biology?

What first intrigued me about computer science was that once you understand the logic of code, everything makes sense. You write a code and enter it into a computer and it will just do whatever the code says. Any errors are part of the code itself. Once I understood it, I found I was really in love with this logic. And in biology, you don't always understand what goes wrong in disease, how mistakes in our genome get translated. All those pieces are missing. But biology also has its own form of logic, and I was curious if I could use the logic in computer science to help me understand the logic in biological science.

Studying only biology limits you to a specific type of research, and the same goes for computer science. While the research within each discipline can be quite distinctive, when you merge these different fields, then you have amazing things like [the AI protein-structure prediction algorithm] AlphaFold. Before, biologists haven't been able to study the function of certain proteins because their structures were unknown. Now, with AlphaFold predictions, biologists have so many new hypotheses.

Being at the leading edge of science that intersects with computer science and biology is super exciting. It opens up new avenues of scientific exploration and pushes the limit of what we are able to do.

How are you using computational methods to study biological systems?

My goal is to build biotechnology tools and computational methods that can enable us to create an in silico cell to simulate interventions of treatments so that we can understand and treat disease. 

Perturbing genes at scale in a dish is one of the easiest and most cost-effective ways to help us understand the functions of genes and find new therapeutic options. The problem we're really facing is that, for example, knocking out a single gene is not enough to cure cancer. We might need to treat with two drugs or perturb two pathways simultaneously to effectively cure cancer. We have 20,000 genes and if we want to exhaustively explore the effects of perturbing any two genes, that's 200 million options. But people have found that even perturbing just two pathways won’t be enough. The problem is just enormous; it's not solvable purely in the lab. However, by understanding how genes interact with each other using existing knowledge in an in silico cell, we can simulate unseen relationships between genes — even if we haven't seen this perturbation, models can learn from the existing data — to predict what we’d see in a dish.

With the help of machine learning, we really scale down the number of experiments that we have to do and can focus on the path that's most promising for therapeutics.

What is it like to juggle computational and wet lab research?

In undergrad I took both biology and computer science courses. But what I found frustrating was that even with experience in both fields it was still hard to communicate to bench scientists because you need to get into really technical details and I didn’t know them. I thus decided to dive more into the wet lab so that I could bridge the two fields and drive effective collaborations. I decided that during my postdoc I wanted to get training in both.

What are you working on now?

In one project, I've been using cell images from Paul Blainey’s lab, where we perturb the genome using a new biotechnology tool that's currently on bioRxiv called CROPseq-multi, which  allows us to look at genetic interactions in the image space in a pooled fashion. That's something that we could never do before because we just didn't have the tools. In the past, scientists have used images mainly in small-scale experiments to validate hypotheses, taking a few under the microscope to see if expected phenotypes show up. But we now finally have the power and the technology to easily generate large image datasets, empowering machine learning to help us understand what changes in cell morphology mean and connect morphology and genetics at a genome-wide scale. 

How could an in silico cell improve the treatment of disease?

I wasn’t always aware of the imbalance in access to medical resources. My grandpa in Rizhao [a city in northeastern China] had cancer and even though I work in the field of cancer research, there was nothing I could do to help him. This made me realize that I'm in a privileged setting where I'm surrounded by medical experts, but the amount of medical resources that he had was totally incomparable to the ones I'm exposed to. That really got me thinking about how we could address such disparities, and one approach could be using an in silico cell to simulate different disease contexts using patient information including genomics and give therapeutic options to patients. With such a model, we could alleviate the inequitable access to medical knowledge for patients around the world. 

We also need new therapeutics and there are so many diseases where we don’t know the cause and don’t even have a therapeutic option available for patients who are suffering. This research can help us find the direct pathway that we should target in personalized genetic contexts. Hopefully this can inspire new therapeutic developments from pharmaceutical companies.