Dr. Mario Roederer
Revolutionizing Flow Cytometry
“As the user drags the mouse over the distribution, FlowJo uses feedback from the mouse to identify the population a user is interested in; this involves an enormous number of computations to determine the optimal gate based on the multidimensional distribution of the cells of interest.
“When users see the region that is close to what they want, they just click and FlowJo creates the auto gate. Before releasing the mouse, the user can optionally fine-tune the gate.
“It’s a sophisticated algorithm,” Roederer continues, “that got tied down to a fairly simple and very intuitive user interface. People just go nuts for that because they think its cool. Scientifically it’s not that important, but from a user interface standpoint, it’s very useful — and in the end, that helps users do good science.”
Biology in QuickTime
FlowJo even makes QuickTime movies.
“It’s so much fun when somebody shows a movie that FlowJo created in a presentation,” Roederer says. “It’s not an analysis tool, but it helps researchers uncover subtle relationships in data that would be impossible to see otherwise.”
Researchers use the movie platform to observe how a distribution of cells changes over a period of time or the distribution of cells in two measurements depends on a third measurement — essentially providing a novel 3-D visualization tool.
But its greatest value involves batch analysis — where users perform repetitive analyses on dozens or even thousands of different samples.
Finding Patterns
“In our field, researchers often have to make minor adjustments to some of the analyses. FlowJo is very good at that, because it allows you to do highly repetitive analysis as well as sample-specific modifications. That’s really its power.”
FlowJo’s movie capabilities give researchers the ability to look at a report in movie form — where every frame of the movie is one sample.
“FlowJo allows you to do highly repetitive analysis as well as sample-specific modifications. That’s really its power.”
“If you’re looking at 200 samples in a movie,” Roederer says, “you can easily catch a mistake, because the brain is very good at picking out patterns and exceptions to patterns. Researchers can stop the movie, step back to a frame and know they have to change the analysis on that one sample.”
Keeping up with the G5
FlowJo’s storage/retrieval process varies for different sites. For example, at the Vaccine Research Center at the National Institutes of Health, data is permanently stored on two Xserve RAID arrays with about four terabytes of data. The facility collects about 30 to 40 gigabytes of data a month — about 20,000 files. All of the data is available online so users can access it any time.
“The data is stored in one safe location,” says Roederer, “and you can analyze it from your computer wherever you are in the world.
“I just had my office hardwired for gigabit ethernet, because my G5 and FlowJo were processing the data faster than it could come across the network. So now I’ve got transfer rates that are 10 times as fast and the disk almost keeps up with the data.”
By Scientists, for Scientists
Roederer believes one reason FlowJo has been so well received is that it was developed by the people who use it rather than by programmers “who sit at a company far away.
“It was an interesting combination of teamwork,” says Roederer. “Adam had extensive experience in user interface and application design and I had extensive experience in the algorithmic and user experience.
“That unique combination made FlowJo work really well. Universally I hear ‘I can’t believe I used [another program] for so long’ from people who get retrained from other programs and start using FlowJo.
“I’ve never met anyone who, after learning how to use FlowJo, went back to a different program.”
