Jean Fan, a PhD student in Medical Sciences studying bioinformatics and integrative genomics, is a statistician, computer scientist, and founder of CuSTEMized, a nonprofit initiative that encourages young girls’ interests in science, technology, engineering, and math (STEM) through personalized books designed to help girls envision themselves as active participants in STEM fields. Fan illustrated and co-wrote The Little Book of Big Dreams, a customizable book made possible by web-based software. This colorful picture book, written in rhyme, recently won a prestigious Moonbeam Spirit Award for encouraging exploration. “When we empower our girls with the confidence and STEM skills to create tools for their community,” Fan says, “they will change the world.”
What led you to Harvard?
I worked as a bench scientist at the National Cancer Institute of the National Institutes of Health before attending Johns Hopkins, where I majored in biomedical engineering and applied math. At NIH, I was working in what is known as a wet lab, working directly with individual breast cancer genes to knock them down—to silence their expression and study the effect of that. By physically modifying a gene’s expression through these experiments, we hoped to glean more insight into the genetic basis of cancer. But this can be very slow going.
I became increasingly interested in computational approaches to cancer research—using genomics, algorithms, computer science, and statistics to analyze which gene is most likely to be driving a particular disease. These data-driven approaches help to prioritize and target research. Now, in my PhD work, I create mathematical computation software for genetic research that allows scientists to explore the data they’ve obtained and helps them further investigate cellular functions.
You’ve taken a keen interest in working with—and now leading—organizations that strengthen the scientific identity of elementary school girls, many of them minority students. What was your first exposure to science?
For the longest time, I wanted to be an artist. I applied to two magnet high schools, one in the arts and the humanities, the other in math, science, and technology. I ended up being admitted to the math and science school, and it was actually my French teacher who encouraged me to attend. It was very challenging at first. When I arrived as a freshman, I didn't have the same STEM background as my peers, and I struggled immensely. After my first class in computer science, I was actually discouraged from continuing.
So what made you stick with it?
On a personal level, I like proving people wrong. I also knew computer science was important. There’s something very special about being able to create a piece of software, share it with anyone, and then they can run it. In a sense, I can’t clone myself, but I can put a lot of brainpower into making a tool that everyone can use, and in a way, replicate my efforts.
What motivated you to found CuSTEMized?
Part of it was to offer elementary-school girls what I would have wished for my younger self—for someone to tell me that it doesn’t matter if you fail here and there. In science, especially, it’s important to get back up and try again. But more immediately, when I started my PhD program in bioinformatics, I was the only woman in my entering class.
This led me to explore more fully why women aren’t coming into these fields. One reason is that there aren’t enough girls developing an interest in STEM fields at a young age—it’s a pipeline issue. That’s when I started volunteering at Science Club for Girls, and I now teach computer science at the Innovation Institute. When I left Science Club, I wanted to give my girls something that would teach them about other STEM fields I didn’t have a chance to cover. Most of them were minority students, and it’s hard to find something that shows them someone in science they can identify with. I started CuSTEMized to address that need.
What made you think that software could help solve a pipeline problem and draw more girls to science?
Our first goal was to create a personalized connection with a young girl to help her envision a place for herself in science—to see that she can be anything she wants to be. We needed something that both cultivated that personal connection and that was scalable. Code is what enables us to generate a custom book easily and immediately. With computer code, you can construct interchangeable pieces to create a character-builder that parents can use to make an avatar for their child, name it after her, and make her the protagonist of the story.
These books are a very personal introduction to a variety of STEM careers K-5 girls might not be aware of. A lot of my students had misconceptions about STEM—they seemed to think that it was all about being in a lab or, surprisingly, working with rats. We want to also teach them about what it’s like to be a programmer or meteorologist or a mathematician. We also emphasize that it is going to be challenging, but at the same time, it’s going to be very interesting and exciting—there’s lots of things to learn and lots of things to explore.
We couldn’t do what we do without software and the Internet. Most of our traffic is online. After parents make the book on our website, they can download it for free and print it at home. We have over 10,000 downloads so far. Parents can also purchase professionally printed hardcover books. We use revenue from these hardcover book sales to run our parallel program that partners with Science Club for Girls to provide personalized books for free to underserved communities. This past year, we worked with the mentors in the program to personalize over 100 books for all their first and second grade students.
How did being a graduate student in bioinformatics influence how you conceptualized CuSTEMized? And how does CuSTEMized contrast with graduate work?
Like my graduate work, CuSTEMized is founded on my interest in using computer science to solve a problem and to create a scalable tool. But the approach to software for CuSTEMized is very different than in bioinformatics. In biomedical research software, there’s a saying that a lot of our software is research grade, meaning that a lot of the code is built for the sake of supporting a publication and it’s not necessarily meant to be reusable. In contrast, for CuSTEMized, I operate much more on what is called an industry standard. We have to make the code reusable, understandable, and it has to be documented.
The nature of the work is also very different in terms of outcome. CuSTEMized, is something that, working with a team, we can push really hard and create a tangible product that people will enjoy. We know where we’re going with it. In research, outcomes are less clear. You can work very hard and sometimes things don’t come together—that’s the nature of research.
Our programming efforts at CuSTEMized tends to be more team oriented and collaborative. In my PhD research, while I collaborate frequently with bench biologists, I typically code alone. But in CuSTEMized, we do a lot of paired programming. We have two people work on exactly the same code for the same project.
What makes a good mentor?
A lot of science is about doubt. In my work, for example, I have to prove with a certain degree of statistical significance that what has been observed is not a chance occurrence. We have to observe a phenomenon a sufficient number of times to rule out that possibility. So there’s always this constant doubting of our results, because it’s part of the nature of what science is.
Sometimes it’s easy to turn that doubt inward instead of engaging with it as an essential feature of the discipline of science. A good mentor is in a way a cheerleader. Scientists tend to be really hard on themselves. It can help to have somebody remind you of your accomplishments and encourage you to continue pursuing an idea and help propel you across the finish line.
Photos by Tony Rinaldo