View Source

t-SNE is a visualization method commonly used analyze single-cell RNA-Seq data. Each cell is shown as a point on the plot and each cell is positioned so that it is close to cells with similar overall gene expression. When working with multiple samples, a t-SNE plot can be drawn for each sample or all samples can be combined into a single plot. Viewing samples individually is the default in Partek Flow because sample to sample variation and outlier samples can obscure cell type differences if all samples are plotted together. However, as you will see in this tutorial, in some data sets, cell type differences can be visualized even when samples are combined.

Using the t-SNE plot, cells can be classified based on clustering results or differences in gene and pathway expression.

Multiple single-sample t-SNE plots

By default, each sample in a multi-sample data set is plotted on its own t-SNE.

Select the Single cell data node
Select t-SNE from the Visualizations section of the task menu (Figure 1)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-28 10:56:20.png

Select Finish from the t-SNE dialog to run t-SNE with the default settings

A t-SNE task node will be generated (Figure 2).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-28 10:58:29.png

Once the t-SNE task has completed, we can view the t-SNE plot.

Select the t-SNE task node
Select Task report from the task menu (Figure 3)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-28 10:59:59.png

The t-SNE plot will open to the first sample in the data set, Astrocytoma 1 (Figure 4). Please note that the appearance of the t-SNE plot will differ each time it is drawn so your t-SNE plots will look different than those shown in this tutorial; however, the cell-to-cell relationships indicated will be the same.

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:42:30.png

The t-SNE plot is in 3D by default. You can rotate the 3D plot by lef-clicking and dragging your mouse. You can zoom in and out using your mouse wheel. The 2D t-SNE is also calculated and you can switch between the 2D and 3D plots using the Plot style radio buttons.

Each sample has its own plot. We can switch between samples using the Back and Next buttons on the upper left.

Select Next

The t-SNE plot has switched to show the next sample, Astrocytoma 2 (Figure 5).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:42:57.png

The goal of this experiment is to compare malignant cells from two different glioma subtypes, astrocytoma and oligodendroglioma. To do this, we need to identify which cells are the malignant cells we want to include and which cell are the normal cells we want to exclude.

The t-SNE plot in Partek Flow offers several options for identifying, selecting, and classifying cells. In this tutorial, we will use expression of known marker genes to identify normal cells.

To visualize expression of a marker gene, we can color cells on the t-SNE plot by their expression level.

Open the Color by drop-down menu
Select Gene expression from the drop-down menu (Figure 6)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:45:17.png

The cells will turn black and a text box Gene ID will open below the drop-down box.

Type CD14 in the Gene ID text box
Select CD14 from the list of genes in the data set (Figure 7)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:44:50.png

The cells will be colored from black to green based on their expression level of CD14, with cells expressing higher levels more green (Figure 8). CD14 is a known marker for microglia and macrophage cells, used by the authors of the original study to classify microglia/macrophage cells.

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:46:8.png

In Partek Flow, we can color cells with up to three genes at a time. We will now add a second gene, MOBP.

Select the icon next to CD14
Type MOBP in the new Gene ID box
Select MOBP from the list of genes in the data set

Cells expressing MOBP are now colored red and cells expressing CD14 are colored green. Cells expressing both genes are colored yellow, while cells expressing neither are colored black (Figure 9).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:46:49.png

Relative expression of the two genes for selected cells can be visualized on the legend.

Activate the 3D lasso tool by selecting
Draw a circle around the cluster of red cells (Figure 10)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:47:16.png

Selected cells are shown in bold and unselected cells are dimmed.

The relative expression of the two genes for the selected cells will be shown on the legend as dots (Figure 11).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:47:38.png

Numerical expression levels for each gene can be viewed for individual cells.

Switch modes by selecting
Select a cell by pointing and clicking

The expression level for that cell is displayed on the legend for each gene (Figure 12).

Deselect the cell by clicking on any black space on the plot

Expression values can also be viewing by selecting Gene Expression from the Label by drop-down menu and mousing over a cell.

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:48:1.png

Now that cells are colored by expression of a microglia/macrophage marker, CD14, and a oligodendrocyte marker, MOBP, we can classify any cell that does not fall into one of these two groups as malignant cells. Because t-SNE groups cells that are similar across the high-dimensional gene expression data, we will consider cells that form a group with CD14 or MOBP-expressing cells as same cell type, even if they do not express the marker gene.

Starting with the Astrocytoma 2 sample, we can classify the malignant cells in each sample.

Activate the 3D lasso tool by selecting
Draw the lasso around the cluster of black cells and click the circle to close the lasso (Figure 13). You may need to switch to selection mode and rotate the 3D plot to select only cells from the black cluster

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:48:35.png

Select Classify selection (Figure 14)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:48:58.png

A dialog to give the classification a name will appear.

Name the classification Malignant
Select Save (Figure 15)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:49:22.png

Once cells have been classified, the classification is added to the Classifications section of the panel. The number of cells belonging to the classiciation is listed; in Astrocytoma 2, there are 253 malignant cells (Figure 15).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:49:51.png

Classifications made on the t-SNE plot are retained as a draft until you exit the t-SNE task report. The Save classifications button runs a task, Classify cells, which generates a new Classified cells data node. In this tutorial, we will classify malignant cells for each sample before we save the classifications, but if necissary, you can run Classify cells for one sample, exit the t-SNE task report, and continue classifying the next sample later starting with the Classified cells data node.

Select Next to move to the next sample, Astrocytoma 3
Rotate the 3D t-SNE plot to allow you to select only cells from the black cluster
Activate the 3D lasso tool by selecting
Draw the lasso around the cluster of black cells and click the circle to close the lasso (Figure 16).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:50:24.png

Select Classify selection
Type Malignant or select Malignant from the prompt (Figure 17)
Select Save

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:50:55.png

Repeat these steps for each of the 5 astrocytoma and 3 oligodendroglioma samples

Once all samples have been classified, it is useful to check the number of cells in each sample assigned to each classification.

Select Classification summary (Figure 18)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:52:10.png

The classifications summary lists every sample, the number of cells in the sample, and the number of cells in each classification (Figure 19).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-28 14:59:0.png

With the malignant cells in every sample classified, it is time to save the classifications.

Select Save classifications
Select Save when asked to confirm

The pipeline view will open and the Classify cells tasks will run, generating a Classified groups data node (Figure 20).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-28 15:7:20.png

Multi-sample t-SNE plot

For some data sets, cell types can be distinguished when all samples can be visualized together on one t-SNE plot. We will use a t-SNE plot of all samples to classify microglia/macrophage and oligodendrocyte cell types.

Select the Single cell data data node
Select t-SNE from the Visualizations section of the task menu
Select Configure on the t-SNE dialog (Figure 21)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 11:5:8.png

Deselect the Split cells by sample option under Misc
Select Apply (Figure 22)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 11:10:22.png

Select Finish to run the t-SNE task

The t-SNE task will be added as a new green layer in the analysis tab (Figure 23).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 11:14:6.png

Once the task has completed, we can view the plot.

Select the green t-SNE plot task node
Select Task Report from the task menu

In the multi-sample t-SNE plot, each cell is initially colored by its sample (Figure 24).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 11:21:7.png

Select 2D from the Plot style section

Viewing the 2D t-SNE plot, while most cells cluster by sample, there are a few clusters with cells from multiple samples (Figure 25).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 11:27:29.png

Using the known maker genes, CD14 and MOBP, we can assess whether these multi-sample clusters belong to our known cell types.

Select Gene expression from the Color by drop-down menu
Type CD14 in the new Gene ID box
Select CD14 from the list of genes in the data set
Select the icon next to CD14
Type MOBP in the new Gene ID box
Select MOBP from the list of genes in the data set

After coloring by CD14, a microglia/macrophage marker, and MOBP, a oligodendrocyte maker, these two cell populations are clearly visible (Figure 26).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 11:32:27.png

Activate the 3D lasso tool by selecting
Draw the lasso around the cluster of red cells and click the circle to close the lasso (Figure 27)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:34:53.png

Select Save classifications

These red cells are MOBP positive, indicating that they are the oligodendrocytes from every sample.

Name the classification Oligodendrocytes
Select Save

To clearly see the CD14 positive population, clear the current selection.

Switch modes by selecting
Deselect by clicking on any black space on the plot

Green CD14 positive cells are the microglia/macrophages from every sample.

Activate the 3D lasso tool by selecting
Draw the lasso around the cluster of green cells and click the circle to close the lasso (Figure 28)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:37:25.png

Select Save classifications
Name the classification Microglia
Select Save
Switch modes by selecting
Deselect by clicking on any black space on the plot

Finally, we will classify all unclassified cells on the plot as the malignant cells from every sample.

Select Classifications from the Color by drop-down menu

Cells are now colored by classification, with Oligodendrocytes in blue, Microglia in red, and unclassified cells in grey.

Activate the 3D lasso tool by selecting
Draw the lasso around the grey cells and click the circle to close the lasso (Figure 29)

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 12:39:23.png

Select Save classifications
Name the classification Malignant
Select Save

With every cell from every sample classified, we can proceed to save classifications.

Select Save classifications
Select Save when asked to confirm

The pipeline view will open and the Classify cells tasks will run, generating a new green-layer Classified groups data node (Figure 30).

Flow Documentation > Classify cells from multiple samples using t-SNE > image2017-12-29 11:45:54.png