Platforms

The Platforms context contains a set of algorithmic or model-based tools.

The Platform Context allows you to run specific native algorithms helpful when performing High dimensional data analysis. After identifying a population of interest, select it in the Population panel of the Analysis tree (Figure 1.1), click on the Platforms icon in the navigation bar (Figure 1.2) and select which platform you would like to run. As of FlowJo v11.2, the algorithm options include t-SNE or UMAP for dimensionality reduction, and X-shift or FlowSOM for clustering (Figure 1.3). The Platform options will open in the Discovery panel. (Figure 1.4). 

Figure 1. Platforms 

There are currently four platforms included:

• t-SNE - Dimensionality reduction - T-Distributed Stochastic Neighbor Embedding with opt-SNE¹.
• UMAP - Dimensionality reduction - Uniform Manifold Approximation and Projection².
• X-Shift - Clustering - fast K-nearest neighbors density estimation³.
• FlowSOM - Clustering - self organizing maps⁴.

Note on  Downsampling in FlowJo v11 Algorithm Platforms:  FlowJo 11 divides the requested total by the number of samples to get a target size for each sample.  In the case that the Total Events number exceeds the possibility of evenly dividing that across samples in the VCP, if some samples are too small, it will max out the smallest sample, subtract that amount from the requested total, and divide what’s left of the total by the number of samples to get a new target size for each sample.  This process repeats until all remaining samples are large enough, or we run out of samples (which is the only case where you end up with fewer total events than requested).  To make the downsampling balanced use a small enough target so that the same number of events can be obtained from each target population.

¹Belkina, A.C., Ciccolella, C.O., Anno, R. et al. Automated optimized parameters for T-distributed stochastic neighbor embedding improve visualization and analysis of large datasets. Nat Commun 10, 5415 (2019). 

²McInnes, Healy. UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction. Feb 2018. arXiv:1802.03426.

³Samusik N, Good Z, Spitzer MH, Davis KL, Nolan GP. Automated mapping of phenotype space with single-cell data. Nat Methods.13(6):493-6 (2016). 

⁴Van Gassen S, Callebaut B, Van Helden MJ, Lambrecht BN, Demeester P, Dhane T, Saeys Y. FlowSOM: Using self-organizing maps for visualization and interpretation of cytometry data. Cytometry A. 2015; 87(7):636-45.