MedImmune Automates and Accelerates Flow Cytometry Data Analysis
Streamline the analysis of flow cytometry data in oncology research
Use MATLAB to create and deploy an automated software solution that applies consistent statistical analysis and provides readable, easily used results
- Analysis times reduced from weeks to minutes
- Developer productivity tripled
- Standardized analysis methods implemented companywide
“I’m much more efficient when I use MATLAB rather than an open-source alternative. In just three weeks I developed and deployed analysis software in MATLAB that has already saved MedImmune researchers thousands of hours of effort.”Dr. Jens Koopmann, MedImmune
Biomedical researchers use flow cytometry and fluorescence-activated cell sorting (FACS) to count, characterize, and segregate live cells in fluid mixtures. For example, FACS helps oncology researchers measure the effect of antibodies and other treatments on cancer cells. Analyzing the data produced by flow cytometers, especially in core labs, clinical research institutes, and other high-volume environments, can be a time-consuming and tedious task that saps researchers’ productivity.
At MedImmune, one of the AstraZeneca group of companies, researchers use FAST FACS software to automate the analysis of flow cytometry data. Developed entirely in MATLAB® in just three weeks, FAST FACS software performs statistical analysis of flow cytometry data and produces easy-to-understand tables and plots that researchers insert directly into their electronic lab notebooks (ELNs).
“The MATLAB based FAST FACS software has revolutionized analysis at MedImmune by eliminating a significant research bottleneck,” says Dr. Jens Koopmann, principal bioinformatics scientist at MedImmune. “Analysis that previously took weeks to perform manually is now completed automatically in minutes. In its first six months of operation, the software has saved our researchers the equivalent of 10 years of work.”
In the past, MedImmune researchers processed flow cytometry results manually using disparate tools, including spreadsheets. This approach was labor-intensive and led to inconsistencies in the way experimental data was analyzed, visualized, and interpreted. The researchers needed an automated way to analyze data from a variety of flow cytometers and perform analysis of variance (ANOVA) to quantify the significance of differences identified in the experimental results.
Koopmann had developed tools similar to FAST FACS in the past using free open-source statistical computing software, but several factors led him to look for an alternative. First, the free software made it difficult to produce well-formatted output, and researchers who used the tools indicated that the results were difficult to read. Second, getting technical support for the free software was a challenge. Third, Koopmann encountered reliability and compatibility issues when using different versions of the software.
MedImmune replaced its manual flow cytometry analysis procedures with an automated workflow based on the FAST FACS software developed in MATLAB.
In this new workflow, researchers add a single column of treatment group information to a spreadsheet of flow cytometry data generated by third-party FlowJo software and then email the updated spreadsheet to an administrative account at MedImmune.
The FAST FACS software scans the email message, opens the spreadsheet, and reads the cytometry data into a MATLAB data structure.
The software then invokes Statistics and Machine Learning Toolbox™ functions to perform ANOVA and other statistical analyses on the data. It applies Tukey's honestly significant difference test to perform a pairwise comparison of treatment effects.
Using the formatting capabilities of MATLAB Report Generator™, the software generates a Microsoft® Word® document that contains a report with a table of contents, hyperlinks, section headers, a summary table, and scatter plots showing ANOVA results as well as median and standard deviation lines. FAST FACS emails this report back to the researcher, who inserts it into an ELN.
The FAST FACS software maintains a database of usage metrics in a MATLAB file. These metrics enable Koopmann to report on the number of analyses performed by date, user, or study, as well as CPU usage and time savings made possible by FAST FACS.
FAST FACS has been used by dozens of researchers in the oncology group at MedImmune on hundreds of clinical and preclinical studies, and has recently been made available to respiratory, cardiovascular, and other groups throughout MedImmune and AstraZeneca.
With this basic framework in place in MATLAB, Koopmann was able to extend the code to other high-throughput experiment platforms in a matter of days. Koopmann has developed several additional automated analysis tools in MATLAB, including one for Meso Scale Discovery® assays and another for microarray analysis.
- Analysis times reduced from weeks to minutes. “Prior to FAST FACS, it took a researcher up to three weeks to analyze flow cytometry data manually,” says Koopmann. “Now, the same analysis is performed automatically in MATLAB and results are delivered in minutes.”
- Developer productivity tripled. “MATLAB enabled me to complete development of a scalable enterprise solution in only three weeks,” says Koopmann. “Compared with freeware available for statistical computing, MATLAB makes me at least three times more productive.”
- Standardized analysis methods implemented worldwide. “In the past, each researcher performed analysis in his or her own way, which made it difficult to compare studies,” notes Koopmann. “With MATLAB and FAST FACS we’ve made a breakthrough in standardizing on how data can be analyzed and presented in both preclinical and clinical settings across both MedImmune and AstraZeneca.”
Jens Koopmann, John Prime, Suzanne Mosely, Amanda Watkins, Nadia Luheshi, Robert Wilkinson, Ronald Herbst, David Fenstermacher, and Mathew Woodwark