When Rani Powers was a junior in high school, her younger sister was hospitalized at Children’s Hospital Colorado with systemic lupus.
“She spent two months hospitalized, some of that time in the ICU,” Rani says. “It was a scary time for her health, but it was also hard on her missing school.”
Because her sister would need to pass her final exams in order to receive credit for classes that year, Rani volunteered to help tutor her in math, Spanish and other subjects. Having previously made mystery video games for her sister loosely based on the Nancy Drew series, Rani used her burgeoning programming skills to make educational worksheets and games to help her sister keep pace. Rani’s sister stabilized (and passed her exams!), and when Rani graduated from high school, she found that she had a tough choice: She had a passion for programming, but had also found a love for medical science.
In college, she chose science, joining the CU Boulder Molecular, Cellular and Developmental Biology program.
“But after undergrad, it was difficult to find entry-level biology positions, and so I worked in software for two years to earn income,” she says. During that time, to keep opportunities open for returning to the field of molecular biology, Rani also worked as a freelance science writer and volunteered in a lab at the National Institute for Standards and Technology (NIST) studying how nanoparticles interact with human cells.
That’s when Rani found out about the Computational Bioscience Program at CU Denver, offered on the Anschutz Medical Campus.
“Science and programming – it would allow me to combine those two things,” Rani says.
The program lets students pick their advisors, and Rani reached out to James Costello, PhD, investigator at CU Cancer Center and assistant professor in the Department of Pharmacology at the CU School of Medicine. Costello’s lab is special, combining lab-based experiments with the development of computational ways of understanding data, mirroring Rani’s own interest in science and programming. While performing experiments using cancer cells and a variety of chemotherapy drugs, Rani started looking at how researchers conceptualize patterns in large data sets. In biology, one of these methods is called Gene Set Enrichment Analysis, or GSEA.
Basically, technology now allows researchers to measure the genes that become more or less active after cells are treated with a drug. But one experiment may identify thousands of these affected genes. What do all these genetic data mean? GSEA lets researchers group these genes, discovering which “families” of genes may be working together, for example, to drive cancer.
“Since the idea was originally published in 2003, GSEA has been cited by more than 15,000 scientific publications,” Rani says.
But GSEA doesn’t always tell the full story – discovering which genes are “turned up” or “turned down” requires comparing genes to some sort of baseline. GSEA essentially uses randomness as the baseline. However, Rani looked at the patterns of genes across hundreds of previously published experiments and found that the baseline wasn’t actually random at all. This led Rani to develop a new model in which results from previous experiments could be used as the baseline. And it let researchers ask a different type of question: How are the results from one experiment similar or different to experiments that were previously published?
“We published a method that was a complementary analysis to say, yes, you can perform this widely used analysis and learn something about the biology, but if you do ours in addition to that, you can gain even more from your one experiment,” she says.
It’s a bit like looking at a picture against a black background instead of a white one. And with that changed perspective, new things emerge. In Rani’s case, changing the background sometimes changes the outcomes, putting new gene “families” at the top of a list of suspects or targets.
Rani applied to present her tool, called GSEA-InContext, to the 2018 conference on Intelligent Systems for Molecular Biology, organized by the International Society for Computational Biology. Not only was her research accepted for a proceedings paper in Bioinformatics and an oral presentation at the conference, it won the Ian Lawson Van Toch Memorial Award for Outstanding Student Paper, the highest award given by the society to student work.
Rani will defend her PhD dissertation in March 2019, after which, “there are lots of options for what’s next,” she says. On one hand, she could continue in academia by accepting an out-of-state postdoctoral position; on the other hand, Rani is considering staying in the Denver area to work in industry, doing biotech and genomics research that she hopes can have a more direct impact on people’s lives.
No matter what she chooses, it will include programming and it will include science. And Rani will be using the skills that she developed as a way to make her sister’s life better in the hospital to help make life better for people whose cures hide deep in the data of their genes.
The code implementing GSEA-InContext is freely available to the public and other researchers on Github here.
The video of Rani Powers receiving the Ian Lawson Van Toch memorial award can be seen here.