In a dot plot colored by channel, each event in a typical biaxial plot is colored by its expression on a third channel, called the coloring channel, or the Z-axis channel. This functionality is available in the Gating Editor and the Illustration Editor. Click the links below to jump to the relevant section.
- Dot plots colored by channel in the Gating Editor
- Dot plots colored by channel in the Illustration Editor
Dimensionality reduction (DR) example of dot plots colored by channel
Dot plots colored by channel have an important role in dimensionality reduction (DR) workflows. When looking at a DR map, coloring by channel allows one to correlate spatially collocated groups of events with biological populations:
(The same sample with viSNE map colored by different channel; this can be applied to any other DR analysis)
Dot plots colored by functional channel
Consider the example below of a simple signaling experiment. The coloring channel is pSTAT3. The two samples are unstimulated (Unstim) and IL10-stimulated PBMCs (Stim). The higher intensity of pSTAT3 in CD3+ T cells, shown in the red-colored population, indicates that T cell signaling is occurring in the sample stimulated with IL10.
(Dot plots colored by channel reveal signaling)
Dot plots colored by channel can help with gating
Coloring dot plots by channel can help identify where to gate, by displaying more information to interpret while drawing the gate. This can especially help in situations where drawing a clean gate is challenging.
In the Gating Editor plot settings box, choose a plot type of Dot – Color by Z-axis Channel. Other important settings are indicated but discussed in the later parts of the article.
(Gating Editor configuration for dot plots colored by channel)
The option for coloring channel appears under the gating plot beneath the X and Y axis controls. Each dot on the plot is colored according to its data value in the coloring channel, for example, pStat3, resulting in the plot above.
The channels that were used for advanced analyses are highlighted in the Gating editor. You will see the icon of each advanced analysis next to the channels that were selected within the analysis.
(An example of UMAP. The channels used for UMAP analysis are highlighted in the Gating editor. All advanced analyses label the channels similarly using the corresponding analysis icon.)
In the Illustration Editor, choose Plot type as Dot, then choose to Color by Z-axis channel. After this, use the selector for Z-axis to define which channel to use for coloring.
To generate a figure that colors plots with a selection of multiple channels, open the Layout settings menu and click Add dimension to add a Channels dimension. Even if you already have Channels selected as a dimension for varying along the X- or Y-axis, you can replace that axis or add another dimension for the Z-axis coloring. In the new Figure dimension, use the dropdown selector to change the axis to Z.
(Layout settings for plots colored by channel)
(Resulting illustration with dot plots colored by selected Z-channels)
Note that you can also change the channel selected on the Z-axis by clicking on the Z-axis channel labeling directly in line with the plots.
Within an advanced analysis experiment such as a DR analysis, clustering analysis, or CITRUS analysis, the channels that were used for clustering during the algorithm setup will be highlighted in the Illustration editor. When you open the Select Z-axis channels window, the channels that were selected for the advanced analysis will be labeled with the analysis name next to the channel name. Below is an example showing the channel labeling for a UMAP analysis. You can use the analysis type as a search term or click on the name of the analysis to easily identify and select the channels used for analysis.
(Channels used for UMAP show UMAP analysis next to the channel name. All other advanced analyses have similar labeling format)
There are two plotting methods for dot plots colored by channel: Binned Pixels and Stacked Dots.
In Binned Pixels mode, collisions between events are binned into single pixels based on the chosen resolution of the plot. Each pixel then reports the median value of the events inside of it on the coloring channel selected for the plot.
The Stacked dot option arranges events on top of each other depending on the order in which they appear in the FCS file. The size of each event in pixels can be adjusted with the Dot size field. This functionality is useful for increasing visual quality of sparse data. No binning is performed with stacked dots, so data may be hidden by overlap in some cases. The Dot shape can be configured as well to be a circle, ring, or square. The ring shape (available only for sizes > 3) is hollow in the middle and can help visualize events that might be overlapping because of stacking. Changing dot shape has little impact unless larger plot resolutions and dot sizes are used.
(Comparison of dot plotting method and shape. All plots are 512-pixel resolution. All stacked dot images are dot size of 7)
The Z-channel range is the minimum and maximum of the color scale that is applied to a dot plot colored by Z-channel, and shown in the color bar to the right of the plot. The color bar is the key for associating a color in the dot plot to a particular value for expression intensity. The minimum and maximum values of the color bar are determined according to the selected method:
- Global (all files): The ungated 2nd and 98th percentile values for the channel in question are calculated for all files in the experiment. The smallest 2nd and largest 98th percentile value are used as the coloring range minimum and maximum for all files in the experiment on this channel.
- Local (per file): The ungated 2nd and 98th percentile values are calculated for the channel in question and assigned as coloring range minimum and maximum on a per-file basis.
- User-defined: The coloring range boundaries are set through user-defined values. One custom range can be defined per illustration and is not channel-specific.
The data values that define coloring behavior and range are calculated on data scaled according to the scaling method of the channel coloring the plot (and compensated, as applicable). However, the data values are untransformed for reporting in the color bar. The coloring follows a linear relationship through the transformed data, but not in the raw data. This is why for non-linearly transformed scales (arcsinh, log) the raw values reported in the color bar do not grow linearly.
The same color palettes available to other plot types in the Cytobank platform are available for dot plots colored by channel. The dropdown menu for palette selection shows a preview of each.
(Palette selection menu)
The recommended color palette for most purposes is Spectrum. This is not be confused with Rainbow, which is similar but is more heavily represented in the greens. Examples of other palettes are shown below.
(viSNE maps colored by channel with different coloring schemes)