Editor’s note: This story was first published on June 19, 2020, based on a preprint. It’s been updated to reflect the paper’s subsequent publication in a peer-reviewed journal.
It can be easy to picture the body’s fight against cancer as little more than a comic book brawl. You have the good guys on one side — go, immune cells! — squaring off against villainous cancer on the other.
In reality, the picture looks like this:
Cancer is maddeningly complex, and its interplay with the immune system involves a huge cast of cells and much chemical chatter. But scientists at Fred Hutchinson Cancer Research Center want to make studying that complex network significantly easier and cheaper.
Meet Infinity Flow. It is free software designed to enhance a popular method of single-cell analysis. The Fred Hutch team described the Infinity Flow method and open-source computational toolkit in a new paper in the journal Science Advances.
Single-cell analysis helps scientists explore the immune system’s vast, decentralized army. These powerful tools help identify which cells are doing what. They are yielding new insights into how the immune system interacts with diseases like cancer — insights that could one day help improve treatments.
Infinity Flow uses machine learning to greatly increase the number of proteins scientists can hunt for on the surface of cells. Those proteins help scientists classify individual cells and identify what they’re doing.
“It’s very hard to measure proteins at the single-cell level,” said Dr. Etienne Becht, the paper’s lead author and a researcher on the teams of Drs. Raphael Gottardo and Evan Newell. “But if you want to study immune cells, you want to measure proteins. And the more the better, because they allow you to define cell types and functions.”
The teams that created the Infinity Flow technique hope it helps accelerate research worldwide to decipher the biology of cancer.
“Infinity Flow removes a lot of the technical hurdles to being able to ask really deep biological questions of a disease system or an organ system,” said Dr. Mark Headley, an immunologist at Fred Hutch and one of the paper’s three senior authors. “I think it will increase the productivity of a lot of labs very quickly.”
Scientists have several different tools to study individual cells. The oldest and most widespread is called flow cytometry.
‘Flow’ refers to how the technique works: Cells flow in a single file past a series of lasers and photo detectors, which read them as they pass. Flow cytometry lets biologists classify millions of cells based on properties like size and the presence of surface proteins.
Immunologists have relied on the technique for years, but it has limits. Scientists can only search for 20 or so protein markers at a time. More recent (and expensive) techniques let researchers hunt for many more surface markers. But they’re in turn limited by the number of cells that can be studied.
“In general, there’s a trade-off between the number of cells and the number of proteins you can measure,” Becht said.
Infinity Flow gets around the trade-off. Here’s a high-level example of how it works:
Scientists use flow cytometry to jointly perform two experiments. In one, they exhaustively study 10 million cells for 20 proteins. In the other, they hunt for 300 proteins in 1/300th of their total cell sample.
The results are then fed into the Infinity Flow software. Using machine learning algorithms, the software predicts the presence of the 300 proteins across every single cell in the experiment. This information lets scientists profile subsets of cells with unprecedented depth.
In their paper, the research team describes how they used Infinity Flow to precisely define the cell populations that compose the lungs of normal healthy mice and leverage that in improving our understanding of how the lung changes in cancer. Headley used Infinity Flow to study a unique and rare population of immune cells called macrophages that “eat” small pieces of tumor cells.
“Macrophages are some of the earliest immune cells to sense the presence of cancer in the lung, and this finding suggests we are only at the beginnings of understanding the diversity present in this population of cells, he said. “Infinity Flow helped us identify these cells and provided us with markers for further study.”
In collaboration with Drs. Peter Morawski and Daniel Campbell at the Benaroya Research Institute, the researchers also conducted a set of tests using donated human blood cells to demonstrate that their method works well, too, in samples with relatively small numbers of cells — often, all that’s available from a patient biopsy. The team gave distinct markers to multiple smaller samples, combined them, and then ran that combined sample through the Infinity Flow assay. They showed they could capture information from hundreds of different proteins on these cells, and tell which original sample each cell came from — making it dramatically easier to analyze smaller sample sizes and interpret results across different patients than with standard methods.
Headley and Becht noted that Infinity Flow uses predictions, so there is a level of uncertainty about the data. “You will still need to go in and validate,” Headley said. “You can’t take this as the final story, and that’s true of every single-cell analysis technique.”
But the upside is huge. Nearly every immunology lab, and a significant number of biology labs, already have flow cytometers. All they need to do is to apply Infinity Flow to turbocharge their research, its creators said.
Just as the Infinity Flow team spans disciplines at Fred Hutch, the software is designed to bridge the worlds of biology and data science.
“We don’t always have extensive training in each other’s respective fields,” said Becht, who is firmly in the computational camp. “We realized it would be more efficient if we could empower biologists. With Infinity Flow, the sky is the limit now for biologists using flow cytometry to study the immune system.”
In fact, said Headley, since the paper’s release in 2020 as a preprint, the research team has heard from scientists around the world who have begun using it in their projects.
"It’s doing what we had hoped it would, which is pretty exciting,” he said.
This research was funded by grants from the Fred Hutch Translational Data Science Integrated Research Center, the National Institute of Allergy and Infectious Diseases., Metavivor, and The Roberta Robinson Fede Endowment.
Jake Siegel is a former staff writer at Fred Hutchinson Cancer Research Center. Previously, he covered health topics at UW Medicine and technology at Microsoft. He has an M.A. from the Missouri School of Journalism.
Are you interested in reprinting or republishing this story? Be our guest! We want to help connect people with the information they need. We just ask that you link back to the original article, preserve the author’s byline and refrain from making edits that alter the original context. Questions? Email us at firstname.lastname@example.org