Skip to main content
AboutGuideContact
Sign in / Register

Flow Advanced Analysis

Dimensionality reduction, clustering, derived parameters, specialized analyses, and MFI comparison.

Dimensionality Reduction

Reduce high-dimensional flow data to 2D for visualization. Three algorithms are available:

  • UMAP preserves both local and global structure and is the best default for most experiments
  • t-SNE emphasizes local cluster separation, useful for finding rare populations
  • PCA is a fast, deterministic linear projection that gives you a good first look

Select the parameters to include, configure algorithm settings, and run. Results appear as new virtual parameters that you can plot and gate on just like any other channel. Multiple algorithms can coexist on the same analysis, so you can run UMAP and t-SNE side by side and compare.

Dimensionality reduction can run locally in the browser or on the server. The toggle is in the run dialog box. Server runs use compute usage; see Flow cytometry compute.

Clustering

Identify cell populations without manual gating:

  • K-Means partitions events into a specified number of clusters quickly
  • DBSCAN finds clusters of arbitrary shape using density and identifies outliers
  • FlowSOM uses self-organizing maps designed specifically for flow cytometry data
  • Phenograph is a graph-based algorithm that auto-detects the number of clusters via KNN + Louvain, useful for high-parameter panels where you don't know population count in advance
  • Density is a 2D-specific clustering method that draws polygon gates around dense regions, ideal for proposing gates on a single scatter plot

Pick the markers relevant to your question, typically 3 to 8 fluorescence channels. More than that adds noise and slows the algorithm, so the parameter selector is capped at 8. If your experiment needs more, get in touch.

Cluster assignments appear as a virtual parameter on your plots, color-coded by cluster. For Density clustering, results land directly as polygon gates in your hierarchy.

Like dimensionality reduction, clustering has a local/server toggle. Server runs use compute usage.

Derived Parameters

Create calculated parameters from formulas based on the channels in your FCS file. Open Derived Parameters from the sidebar to build expressions like:

  • CD4 - CD8 for the difference between two markers
  • log(FSC-A * SSC-A) for a derived scatter metric
  • (CD25 + CD127) / 2 for averaged signal

Derived parameters appear in the parameter picker for every plot, gate, and statistics table. They recompute automatically when the underlying channels change.

Specialized Analyses

Cell Cycle

Quantify the distribution of cells across cell-cycle phases (G0/G1, S, G2/M) from a DNA content histogram. The Cell Cycle dialog box auto-detects common DNA stains (DAPI, PI, Hoechst, DRAQ, 7-AAD) and fits a Dean-Jett-Fox model. Output is phase gates with percentages.

Proliferation

Analyze cell division from dye-dilution assays (CFSE, CellTrace Violet, CellTrace Far Red). Auto-detects the dye parameter and produces generation peaks plus division and proliferation indices. Configure max generations (default 8) and the event cap.

Kinetics

Find time-dependent artifacts and changes in your data. Useful for calcium-flux experiments, signaling time courses, and acquisition stability. Auto-detects the TIME parameter, bins it, flags anomalous bins, and suggests time-based exclusion regions.

Quality Control

QC runs a multi-parameter time-gating pass across every fluorescence channel at once. It surfaces a stability score, a grid of mini time-plots with anomalies highlighted, and suggested exclusion regions you can apply with checkboxes. See Flow files and templates for the full QC workflow.

Correlation Heatmap

Open Correlation Heatmap to visualize pairwise Pearson correlation between parameters. Useful for spotting redundant channels or finding unexpected co-variation between markers.

MFI Comparison

The MFI Comparison dialog box compares geometric mean fluorescence intensity across markers between a control and a sample population. Requires at least two gated populations.

Select a control and a sample from the dropdowns, then review the results table:

  • Control and sample MFI values
  • Fold change and log₂(FC) for each marker
  • A divergence bar chart (blue for upregulated, red for downregulated)
  • Kolmogorov–Smirnov p-values for statistical significance

Sort by any column and export as CSV. Significant rows (p < 0.05) appear in bold. Scatter, side-scatter, and TIME parameters are filtered out automatically.

Compare Files

The Compare Files dialog box runs a side-by-side comparison across multiple FCS files in your working set. Useful for batch-vs-batch QA and treatment-vs-control comparisons across cohorts.

Next Steps

Flow visualizationFlow files and templates