Massachusetts Academy of Math and Science
Data collected from different sources may have substantive differences, even if the underlying experiments are identical. These differences are known as batch effects. Removing batch effects is critical for being able to compare biological data collected under different circumstances. Because they often look similar to biological effects at first glance, batch effects can be difficult to identify, let alone remove. Andrew developed a new computational approach to reduce batch effects using principal component analysis. He was able to show that when applied to real single-cell RNA-seq data, his algorithm performs favorably compared to the mutual nearest neighbors batch-effect-correction method.
“My favorite part about working at the Broad this summer was by far interacting with other scientists and researchers who think just like me,” said Andrew. “I find it amazing how the Broad brings together people from all different walks of science to solve a large medical issue. This interdisciplinary background (and the openness of the facility itself) help to foster a truly creative and driven workspace that I had never before experienced.“