Cytometry — encompassing spectral and conventional cytometry and mass cytometry — is unquestionably the premiere single-cell profiling assay technology for measuring and gaining insights into the immune system. It is conceptually simple, high-throughput (capable of measuring millions of cells per sample at low cost), and — with the advent of mass and spectral — high-dimensional. It is widely used across academia and industry to characterize the human immune system in research, pre-clinical and clinical trials, and drives modern drug development forward.
Yet for a technology that has been in practice for over half a century, it continues to face technical challenges — panel development, analytics, experimental practice and standardization, and others. This blog series will discuss the challenges inherent in flow cytometry, how Ozette has overcome them, and hopefully provide readers with information and insights to help them generate higher quality data, more reliable biological insights, and an understanding of state-of-the-art tooling and workflows.
For decades, flow cytometry has promised immunological insights and cellular biomarkers via deep immunological profiling. A typical clinical trial can span many months or years. Yet outside of a few highly specialized labs, generating high quality flow cytometry data reliably over time is surprisingly challenging — so much so that one of the key mandates of ISAC (the International Society for Advancement of Cytometry) is education and training. Many practitioners struggle with all aspects of the assay: panel design, instrument calibration and standardization, generation of appropriate controls, unmixing or compensation, and downstream analysis.
The apparent simplicity of the technology masks layered complexity. Physicists and engineers design the instruments, chemists design the reagents, immunologists design the panels, and computational biologists develop methods to process data. The field is exceedingly interdisciplinary — which is what makes it exciting. With all this layered complexity, it is unlikely that any one individual would be an expert in all areas.
Yet when it comes time to generate the data, there is often a singular individual who stains the cells, operates the instrument, acquires the data, compensates/unmixes, processes and gates the data, and analyzes the results. If something in the multi-layered cytometry "cake" is off, and the person responsible is not sufficiently expert in nor equipped with the necessary tools, what happens? In our experience, data quality suffers and experiments fail, the reasons "unknown or unexplained," and consequently flow cytometry gets a reputation as a fickle technology.
In contrast to software solutions that push data quality monitoring onto an ill-equipped individual, Ozette proactively takes ownership of data quality, recognizing that robust biology begins and ends with high quality data. We have built tools to diagnose and remediate data quality issues, and developed significant internal expertise, generating high quality data in our GCLP lab. Through this blog we will point readers to some of the lesser-known pitfalls, and convey some of the expertise of our multidisciplinary team.
This blog will cover Ozette's opinionated view of the practice of flow cytometry — technical and non-technical aspects of data generation and analysis, including control generation, unmixing and compensation, processing, gating, and downstream analysis and inference, coupled with some case studies. We will discuss tooling and methods, their appropriate and (in our opinion) inappropriate uses, organizational efficiency, collaboration in multidisciplinary teams, and the importance of transparency, observability, interpretability, and "explorability" of complex data.
To date, we have worked with customers spanning pharma, CRO, academic, and core labs. We have seen a lot of data, and identified common, core areas that underlie the most common technical failings. If you struggle with or worry about flow data quality, timeliness of results, biological reproducibility, or obtaining reliable insights for decision making, then this blog will be of interest to you — whether you generate, process, or consume the data (or all of the above).
← All posts