Use Case 1: Load Data via scNetViz Browser

• Launch Cytoscape. In the Apps menu, select scNetViz → Load Experiment → From Single Cell Expression Atlas..., or click the icon in the Cytoscape Toolbar.
• In the Single Cell Experiment Atlas (SCEA), click on the column header for the Accession column to sort it.
• Locate the experiment with acession number E-GEOD-81383 and click on the row to highlight it. Click View Data to load the data.

Calculate Differential Expression

The data opens the Experiment Data Table, which contains normalized scRNA-seq quantification in transcripts per million in the TPM tab, metadata like experimental variables etc in the Categories tab, and for SCEA it also includes the grouping of cells as calculated by unsupervised clustering. The results displayed are for different values of the K parameter.

• In the Categories Tab locate the clustering result with a sel.K value of True. This represents the best clustering result for the data.
• Click the Calculate Diff Exp button to calculate differential expression.

Create Networks

The DiffExp tab will now be populated with data for the chosen cluster result.

• Click the Create Networks button to create networks. This will create networks by querying the STRING database for the protein products of genes that satisfy the Network Analysis cutoffs based on FDR and Log2FC, or Max genes count.

Create Networks

In this case, 6 networks are created; one for each cluster and one for all the clusters combined. The Log 2 Fold Change for each cluster is visualized as the node fill color. In the Results Panel, two tabs will be loaded, for scNetViz and stringApp.

Functional Enrichment Analysis

• Click on one of the clusters in the Network Panel to select it.
• In the Get Enrichment section of the ScNetViz tab, leave the default selections as-is and click Retrieve Table.
• The functional enrichment results will be loaded in a new STRING Enrichment panel in the Table Panel below the network.
• Click the Draw charts using default color palette button .
• Repeat these steps for the remaining networks.

Compare and Contrast

• Explore the 5 networks. Note that this data set includes three melanoma cell lines.
  • Do the enriched terms make sense?
  • Do you see any differences between the networks?
  • Do you see any similarities between the networks?
  • Look at the enriched terms for clusters 2,3, and 4 and compare with clusters 1 and 5. See any consistencies or differences?

Use case 2: Load a Local scRNA-Seq Dataset

In this workflow, we will load normalized data and cluster assignments from local files.

• Download the normalized data (zipped folder).
• Download the cluster assignments:
  • On Chrome/Firefox: Click on the link above, and when it opens in the browser, right-click and select File → Save Page As....
  • On Safari: Right-click the link above and select Download Linked File As....

Load data

• Launch Cytoscape. In the Apps menu, select scNetViz → Load Experiment → Import from file....
• Select Homo sapiens as the species, by typing directly into the drop-down. Browse to find the E-GEOD-109979 zip file under File or directory with MTX matrix and headers. Click OK to load.
• Leave Skip first line of header files blank.

Load Categories

• In the Experiment Data Table, click the Add Category drop-down and select Import from file....
• Browse to select the .csv file with category data, and leave remaining selections as the default.

Load Categories

Once the categories are loaded, the data viewer is updated:

Create UMAP Plot

• Under New Cell Plot, select UMAP. Keep the default parameters.

Calculate Differential Expression

• In the Categories Tab locate the clustering result with a sel.K value of True. This represents the best clustering result for the data.
• Click the Calculate Diff Exp button to calculate differential expression.

Create heatmap

Next, we will generate a heatmap showing the top differentially expressed genes:

• Under View Plots, select Heatmap.

Create Networks

• In the DiffExp tab, click the Create Networks button to create networks from the STRING database.

In this case, 5 networks are created; one for each cluster and one for all the clusters combined. The Log 2 Fold Change for each cluster is visualized as the node fill color.