Clustered t-SNE analysis

Enables improved visualization of t-SNE plot trends through algorithmic clustering

This tab displays the t-SNE plot after applying a clustering algorithm (selected in the setup, e.g., Louvain) to the 2D t-SNE coordinates. This helps to automatically identify and label distinct groups within the visualization.

Screenshot of the Clustered t-SNE Analysis tab showing colored clusters

What it Shows

• Clustered t-SNE Plot: The main visualization is the standard t-SNE scatter plot, but now the points (samples) are colored according to the cluster they were assigned to by the chosen algorithm.

• Cluster Labels: Each distinct cluster is typically labeled directly on the plot (e.g., "1 - 54" indicates Cluster 1 containing 54 points).

• Legend: A legend maps the colors back to the cluster IDs.

• Algorithm Information: Text above the plot often specifies the clustering method used (e.g., Louvain) and may mention specific details, such as how outliers are handled (e.g., designated cluster "100").

• Silhouette Score: An average silhouette score might be displayed, providing a metric for how well-separated the clusters are (values range from -1 to +1, higher values indicate better clustering).

How to Interpret

• Identify Groups: Use the colors and labels to clearly see the distinct groups identified by the clustering algorithm within the t-SNE map.

• Evaluate Clustering: Compare the automatically identified clusters to any visual patterns you observed in the standard t-SNE plot or based on known grouping variables. The silhouette score gives a quantitative measure of cluster separation.

• Foundation for Further Analysis: These cluster assignments form the basis for the analyses performed in the Dataset Analysis tab, which examine the characteristics of these groups using the original high-dimensional data.

1. Clusters

This view displays the t-SNE plot where samples are colored based on the clusters identified by the algorithm you selected in the setup (e.g., Louvain, Hierarchical).

• Visualization: Each point represents a sample, colored according to its assigned cluster ID. This helps you clearly visualize the groups found by the clustering method within the t-SNE map.

• Silhouette Score: A description above or near the plot usually includes the average silhouette score for the clustering result.

  • This score measures how similar each sample is to its own cluster compared to other clusters.

  • Values range from -1 to +1.

  • A high value (closer to +1) indicates that samples are well-matched to their own cluster and poorly matched to neighboring clusters, suggesting well-defined, distinct clusters.

  • Values near 0 indicate overlapping clusters.

  • Negative values generally indicate that samples might have been assigned to the wrong cluster.

• Download: You can download this plot as an SVG file or right-click to save it as a PNG directly from PANDORA.

2. Features

This view helps you understand the characteristics of the clusters identified in the Clustered t-SNE Analysis by examining the original data features (variables).

• Visualization: Displays a plot (often a grouped bar plot) titled "Mean Values of Features by Cluster".

  • The X-axis represents the different clusters identified by the clustering algorithm.

  • The Y-axis represents the mean value of each feature within a specific cluster (often after scaling/centering if applied during preprocessing).

  • Within each cluster on the X-axis, there are multiple bars, each representing a different original feature, colored according to the legend.

• Interpretation: Compare the heights of the bars for the same feature (same color) across different clusters.

  • Features with significantly higher or lower mean values in one cluster compared to others are potential biomarkers or distinguishing characteristics of that cluster.

  • For example, if the pink bar (representing nasal_foldchange_go.0070498) is much higher in Cluster 1 than in Clusters 2 and 3, it suggests this feature has a high average value specifically in the samples belonging to Cluster 1.

3. FoldChange

This view allows you to compare the relative levels of original features across the different clusters identified in the Clustered t-SNE Analysis, typically represented as fold changes.

• Visualization: Displays a plot, often titled "Fold Change of Features Across Clusters".

  • The Y-axis lists the identified clusters.

  • The X-axis represents the fold change value (often on a log scale).

  • For each cluster, a bar is shown composed of colored segments. Each segment corresponds to an original feature, with its color defined by the legend.

• Interpretation:

  • The position of a colored segment along the X-axis for a given cluster indicates the fold change of that feature in that cluster compared to a baseline (e.g., compared to the average across all other clusters).

  • Segments extending to the right (positive fold change) indicate features that have higher average values in that specific cluster compared to the baseline.

  • Segments extending to the left (negative fold change) indicate features that have lower average values in that specific cluster.

  • The magnitude of the fold change is represented by how far the segment extends along the X-axis.

Plot showing fold change of features across clusters

• Use: Identify features that are significantly enriched (up-regulated) or depleted (down-regulated) within each specific cluster, providing insights into the biological characteristics that define each group.