Pipelines in FCS Express are a set of data processing steps that stand alone or are connected in series.  The output of a step can be applied to a data file or utilized as the input of the next step, or series of steps, that may be applied to your data. No plugins, no programming, just results!

The vast majority of commonly used algorithms for data analysis are actually pipelines, with tSNE (and its variants) and SPADE being two examples. However, when those algorithms are implemented as a one calculation, user customizations are limited to the what the specific implementation allows. Pipelines in FCS Express increase the computational flexibility and granularity of running algorithms and data transformations while giving users the unique ability to create their own transformations.

FCS Express provides a wide variety of pipeline steps for you to choose from. See our full list of all available steps and contact us if there are steps you are interested incorporating into FCS Express that are not yet available.

FlowSOM is a clustering and visualization tools that facilitate the analysis of high-dimensional data. It clusters the input dataset using a Self-Organizing Map (SOM)* allowing users to cluster large multi-dimensional data sets in a short time.

FlowSOM also performs a second clustering step (called meta-clustering) in which clusters, not events, are clustered together based on their similarity. Meta-clustering is a very useful step to group similar clusters together and find meta-clusters which better represent the biological populations in the dataset.

FCS Express 7 allows the user to easily create and run a FlowSOM transformation through the pre-defined pipeline algorithms.

* FlowSOM in FCS Express 7 (>7.06.0015) automatically clusters data using an improved version of the SOM algorithm called Batch-SOM.

Phenograph is a clustering tool that facilitates the analysis of high-dimensional data. Phenograph first represents the high-dimensional data set with a graph structure. In this representation, each cell is represented as a node and is connected to its neighbors (i.e. the cells most similar to it) via an edge whose weight is set by the similarity between cells. Dense regions of cells will manifest as highly interconnected (i.e. high-density of edges) modules in this graph. Once constructed, the graph is partitioned into subsets of densely interconnected modules, called communities, which represent distinct phenotypic subpopulations.

FCS Express 7 can natively perform dimensionality reduction using the UMAP (Uniform Manifold Approximation and Projection) algorithm.

Together with tSNE/viSNE, UMAP is one of the dimensionality reduction algorithms mostly used nowadays.

In FCS Express 7, UMAP can be run as a standalone algorithm or combined with other algorithms (e.g. other dimensionality reduction algorithms or clustering algorithms), or mathematical calculations (e.g. scaling, normalization,...), in more complex pipelines.

FlowAI is a data-cleaning algorithm able to perform three quality controls on your flow cytometry data:

1. Flow rate QC
2. Signal acquisition QC
3. Dynamic range QC

FCS Express 7 allows to natively run FlowAI via the pipeline transformation without the need of plug-ins or external software such as R.

FlowCut allows the user to perform quality control on flow cytometry data in order to improve both manual and automated downstream analysis.

FlowCut is a two step process: It first removes acquisition regions with fewer events and then cleans the remaining data based on multiple quality control tests. To perform the second step, FlowCut separates events along the time axis into equally sized segments and calculates if any of the segments should be removed because they are statistically different than the others.

In FCS Express, the FlowCut pre-defined algorithm automatically creates a Pipeline Folder that includes the two steps above, allowing users to easily run FlowCut on the desired input parameters and events.

The dimensionality-reduction algorithm, t-SNE (viSNE), is available natively in FCS Express to perform and analyze data in real-time.  In the example below, three different files are visualized for four different markers in representative t-SNE maps by quickly moving the blue gate from sample to sample.

In FCS Express, SPADE is executed natively to perform rapid analysis and visualitzation of results.  In the example below, the red gate from the left plot is applied to the SPADE tree (heat map on right) displaying the same data file.  When the gate is moved, the SPADE tree's clusters update automatically in real-time.

The Python Transformation pipeline step allows FCS Express and Python to communicate with each other through a text script that can be created directly within FCS Express.

The Python transformation step allow to integrate popular algorithms such as PARC, TriMap, Ivis... and many other.

R integration allows for new parameters to be added or clustering performed on data files directly within FCS Express.  The FlowClean R script has been executed for the data file below to display good versus bad events as a new parameter (left - logarithmic scale, right - linear scale).

K-means is one of clustering tools in FCS Express.  This partitioning-based algorithm has been performed on a single file below to generate four distinct clusters (heatmap on the right and backgated onto traditional 2D plot on left).

Principal Component Analysis can be performed in FCS Express to create as many or few principal components needed to graph and analyze.

Parameter math can define new or replace parameters by using the existing parameters and constants in sequences composed of simple or advanced mathematical functions.