flowCore: a Bioconductor package for high throughput flow cytometry.

Hahne F, LeMeur N, Brinkman RR, Ellis B, Haaland P, Sarkar D, Spidlen J, Strain E, Gentleman R. · 2009

Read flowCore: a Bioconductor package for high throughput flow cytometry. by Hahne F, LeMeur N, Brinkman RR, Ellis B, Haaland P, Sarkar D, Spidlen J, Strain E, Gentleman R. (2009) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc2684747.txt.

Introduction

Hahne and co-authors released in 2009 flowCore, a software package for handling flow cytometry data at modern scale. Flow cytometry reads single cells as they stream past lasers, a workhorse method in immunology and drug trials. Automation had swollen experiments to thousands of samples, but analysis tools still assumed small, hand-run studies.

Built for the R language inside the Bioconductor project, flowCore gives researchers shared data structures that hold whole cohorts at once. The same steps can then run across every sample in a study, instead of one file at a time. That shared frame lets statisticians, programmers, and clinicians build on each other's work rather than start over. Several later packages already stand on it. The authors argue that common plumbing, unglamorous as it sounds, is what lets a field model rich trial data properly.

Catalog record

CREATORHahne F, LeMeur N, Brinkman RR, Ellis B, Haaland P, Sarkar D, Spidlen J, Strain E, Gentleman R.
YEAR2009
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC2684747

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

flowCore: a Bioconductor package for high throughput flow cytometry By Florian Hahne, Nolwenn LeMeur, Ryan R Brinkman, Byron Ellis, Perry Haaland, Deepayan Sarkar, Josef Spidlen, Errol Strain, Robert Gentleman Published in: BMC Bioinformatics DOI: 10.1186/1471-2105-10-106 ABSTRACT Background Recent advances in automation technologies have enabled the use of flow cytometry for high throughput screening, generating large complex data sets often in clinical trials or drug discovery settings. However, data management and data analysis methods have not advanced sufficiently far from the initial small-scale studies to support modeling in the presence of multiple covariates. Results We developed a set of flexible open source computational tools in the R package flowCore to facilitate the analysis of these complex data. A key component of which is having suitable data structures that support the application of similar operations to a collection of samples or a clinical cohort. In addition, our software constitutes a shared and extensible research platform that enables collaboration between bioinformaticians, computer scientists, statisticians, biologists and clinicians. This platform will foster the development of novel analytic methods for flow cytometry. Conclusion The software has been applied in the analysis of various data sets and its data structures have proven to be highly efficient in capturing and organizing the analytic work flow. Finally, a number of additional Bioconductor packages successfully build on the infrastructure provided by flowCore, open new avenues for flow data analysis. BACKGROUND Automation technologies developed during the last several years have enabled the use of flow cytometry (FCM) to generate large, complex data sets in both basic and clinical research applications [1]. A serious bottleneck in the interpretation of existing studies and the application of high throughput FCM to even larger, more complex problems is that data management and data analysis methods have not advanced sufficiently far from the methods developed for applications of FCM to small-scale, tube-based studies [2]. In particular, the data often need to be organized into groups of samples based on combinations of additional covariates and similar operations need to be applied to these groups in a transparent and reproducible manner. Furthermore, the growing depth of knowledge in the field of immunology, for instance the characterization of distinct human T-cell sub-population [3], clearly argues for more systematic approaches. Some of the consequences of the lag of efficient software solutions are difficulties in maintaining the integrity and documentation of large data sets, assessing measurement quality, developing validated assays, controlling the accuracy of gating techniques, automating complex gating strategies, and aggregating statistical results across large study sets for further analysis. In addition, new analysis approaches face difficulty in finding their way into standard practice. We believe that these barriers to the development and dissemination of new analysis methods is one of the fundamental restraints on the future expansion of FC-HCS in both clinical and research applications. Traditionally, for the majority, FCM experiments were being analyzed by manual data inspection in one or two dimensions, or by very basic comparisons of summary statistics. Most of the currently available analysis tools are designed to reflect this work flow. We believe that these approaches, in addition to being expensive and labor intensive, do not fully address the highly complex nature of FCM data; in particular, they disregard many of the fundamental aspects of the data, such as sample groups or cohorts, the underlying distribution or its high-dimensional nature. Furthermore, the subjective character of manual analyses are a major obstacle to reproducibility. In a recent study of flow cytometric standardization involving 15 institutions, the mean inter-laboratory coefficient of variation ranged from 17 to 44%, even though preparation was standardized and performed using the same samples and reagents at each site [4]. For FC-HCS data, unassisted manual inspection is extremely time consuming, and robust statistical methods need to be developed to point investigators to interesting aspects of the data, or to potential problems. While the expert knowledge of immunologists and researchers remains crucial for the understanding of FCM data, we believe that collaboration with other research fields such as statistics and computer science can greatly improve the relevance of FCM in today's high-throughput paradigm. In this paper, we describe a set of flexible and well structured computational tools to efficiently analyze FC-HCS data. Our intent is to provide a shared research platform that enables bioinformaticians, computer scientists, and statisticians to work collaboratively with biologists and clinicians to develop novel methods for FCM data analysis, a process deemed crucial by many for the further development of the technology [5]. IMPLEMENTATION The computational tools we have developed are distributed in the R software language [6] as the Bioconductor [7] package flowCore. The package flowCore is a freely available, highly functional, and extensible FCM data analysis platform that enables researchers to efficiently handle FC-HCS data and encourages open development of tools for their coherent analysis. In our implementation of flowCore we rely on two important lessons learned from the field of gene expression data analysis: the first being the importance of data structures that reflect the underlying data and facilitate the manipulations that are of most interest, while the second is the importance of a modular architecture that allows for many developers to extend and use the underlying infrastructure and to combine tools in complex work flows. flowCore implements such computationally efficient data structures and a range of specialized methods addressing all components of a typical FCM analysis work flow, including compensation, transformation, and gating. flowCore runs on Windows, Mac OS X, and Linux/Unix operating systems. Existing data standards and conventions Currently, data from FCM experiments are stored in single files according to the Flow Cytometry Standard (FCS) [8]. However, recent developments in high-throughput FCM are shifting the focus of interest away from single-tube based measurements towards large and complex experimental designs with dozens of covariates and influencing factors. For example, experiments consist of large numbers of samples from different patients, measured at different time points [1] or following different drug treatments [9]. Modern FCM data analysis tools have to deal with an additional layer of sample metadata and they need to provide infrastructure to process and to compare groups of samples in a concise and coordinated manner. The notion of classes from an object-oriented programming language provides one coherent way to describe these richer data structures. In addition, functions or methods that work on those classes allow for interaction and manipulation. Many of the currently available software solutions offer only limited support for such self-contained structures, or make use of binary storage containers that are designed specifically for the needs of particular user interfaces and hence are not easily amenable to programmatical access. In addition, the closed-source nature of these products often makes them impractical to integrate into analysis pipelines. In this manuscript we describe classes for FCM data analysis and their implementation in R, however, they could just as easily be implemented in any other language (e.g., Java, C++). Software written in those languages could use similar data structures, thereby simplifying communication and the interchange of data between analysis tools. flowCore does not provide a graphical user interface and all operations are done using a command line interface. It is possible to add a more elaborate user interface on top of this infrastructure, however the

Read the complete text