...

• Click the Normalized counts node 
• Click Exploratory analysis in the task menu
• Click PCA

You can choose Features contribute contribute equally to standardize the genes prior to PCA or allow more variable genes to have a larger effect on the PCA by choosing by variance. By default, we take variance into account and focus on the most variable genes. 

...

• Click Configure to access the advanced settings
• Click Generate PC quality measures (Figure 5)
  
  

This will generate a Scree plot and PC component loadings table, which are useful for determining how many PCs to use for downstream analysis tasks. 

...

•  Mouse over the Scree plot to identify the point where additional PCs offer little additional information (Figure 6)
  
  

...

In this case, we see a few known marker genes are highly correlated with PC1 (Figure 7). 

Graph-based clustering

Graph-based clustering identifies groups of similar cells using PC values as the input. By including only the most informative PCs, noise in the data set is excluded, improving the results of clustering. 

...