In a Starzl Tower lab, Theresa White examines a selection of dyes known in the lab as fluorescence tags. She passes on Texas Red and Zombie NIR, then adds Alexa to her sample solution with a dropper. “Zombie NIR makes the dead cells glow,” she explains.
White fits the sample tube into a sheath on a printer-sized machine called a flow cytometer, an advanced Cytek Aurora model. A needle reaches into the tube like a feeding mosquito and draws up a droplet.
White, a graduate student in the lab of immunology professor Penelope Morel, MD, prepared the sample using fresh spleen cells of a mouse carrying a mutated RNA binding protein, which causes variation in T-cells. The variation should be present throughout the sample, creating myriad data points, because the mutated protein is expressed in multiple cell types.
Flow cytometry offers great possibilities. Cells in solution are tagged with light-emitting dyes – molecules that attach to biomolecules such as proteins – then exposed to lasers that stimulate the dyes, producing a spectrum of color emissions highlighting specific regions of the cell, and individual proteins within those regions. But overlapping wavelengths make it impossible to distinguish among similar colors within the limitations of human vision and of previous technologies.
The computational capacity of the Aurora model removes these limitations. It can read 48 individual color tags on a cell, more than twice the capacity of existing cytometers. The Aurora distinguishes and unmixes similar colors that overlap in the spectrum using algorithms that cluster cells displaying similar color patterns, then identifies individual cells within those patterns at a rate of thousands of cells per second. Data pipelines automate the analysis of files containing data on millions of cells per sample and hundreds of samples per analysis, making it possible to identify and compare an unprecedented range of disease indicators.
Computational flow cytometry has become an important focus of the Department of Immunology, which created the Unified Flow Core to oversee flow cytometry for the 300-plus Pitt researchers using the technology. They represent roughly $150 million in NIH grants.
The Unified Flow Core acquired the Cytek Aurora in March 2018, helped by instrumentation funds from the Department of Immunology and a research project grant from The Clinical and Translational Science Institute (CTSI). Lisa Borghesi, PhD, associate professor of immunology and scientific director of the Unified Flow Core, led the effort.
“The Aurora helps us organize our understanding of the immune system at a level not before possible,” Borghesi explains. “In the past, we analyzed cells by plotting pairs of colors, such as green on the X-axis and pink on the Y-axis. With 18 colors, that scales to 153 two-dimensional plots. The human brain can’t synthesize millions of cells marked by a rainbow of colors into a meaningful interpretation, but computational flow cytometry can.”
Pitt Center for Research Computing partnered with Borghesi to create the computational infrastructure for the data analysis. Central to the project were Pitt CRC consultants Kim Wong, research associate professor, and Ketan Maheshwari, research assistant professor (now at Oak Ridge National Laboratory).
Borghesi describes Pitt CRC as “indispensable.”
“We need the power of CRC’s high-performance computing clusters, but that power needs to harmonize with the graphic user interface tools we use. Pitt CRC helps you move seamlessly from your laptop into the cluster.”
“The software building blocks already exist,” Wong explains. “We create simple-to-use workflows that take advantage of Pitt CRC’s advanced computing. We do the necessary code adaptation, troubleshooting, and user support so that biologists can just focus on solving big problems and not worry about the computer.”
“I love this machine,” White says, preparing fresh samples for the Aurora. “The old machine accommodated five sample panels with eight to 10 colors each. The Aurora does the same analysis with two panels with 18 colors each. You can overlay images of the normal and the mutant regulatory T-cells. Because you sample more events, you get a larger population. You are not trying to make conclusions based on a small sample. Aurora enables a global view – it shows the forest.”
For information on workshops and training resources, go to:
Pitt Center for Research Computing