Cytometry analysis pipeline for large and complex datasets (CAPX)
Analysis pipeline for cytometry data
Sydney Cytometry recommend an analysis approach using a number of existing tools.
1. Data preparation. Data from flow or mass cytometry experiments can be processed initially to select a population of interest (POI) in a program such as FlowJo.
2. Clustering. Once a POI has been exported, large datasets of tens of millions of cells can be clustered using FlowSOM (reference: https://www.ncbi.nlm.nih.gov/pubmed/25573116).
3. Dimensionality reduction. Once clustered, data can be subsampled, and then displayed on a 2D single-cell plot using tSNE (link: https://lvdmaaten.github.io/tsne). Individual 'tSNE plots' can then be generated, where each cell is coloured by the level of expression of various markers (https://github.com/sydneycytometry/tSNEplots). This allows the user greater ease in identify various cell types represented by the FlowSOM clusters.
4. Data exploration. Clustered data and/or dimensionality reduced data can then be exported, and examined in a direct fashion using programs such as FlowJo.
5. Heatmap and point graph summaries. The cluster generated by FlowSOM can then be compared between each other, and across different samples using heatmaps or point graphs (https://github.com/sydneycytometry/autograph).
R script and protocols
A full script in R and usage protocol will be available soon at https://github.com/sydneycytometry/CAPX. Additionally, an alternative protocol for using this approach in the cytofkit package (https://github.com/JinmiaoChenLab/cytofkit) or FlowJo (https://sydneycytometry.org.au/computational-analysis) will be available soon here.
An introductory tutorial to using R and RStudio can be found at this link.