A visual analytics framework for visualisation and analytics of single cell RNA-Seq time course data
Tangerine is a Python based visualisation and analytics framework for single cell RNA-Seq time course data. It allows the user to interactively explore transcriptional programs across the time series.
The pipeline is a two step process - first, data is processed and results are written to disk. Next, these results are visualised in a browser. Here, we mainly look at transcription factor (TF) correlations and TF-gene interactions.
There are three main views in the visualisation -
-
Global topology and module evolution: A global TF correlation heatmap is linked to an alluvial diagram . This allows users to visually select specific TF subsets in a given timepoint and trace their module membership across the time-course. Further, the selected TF subset can be compared in detail across user specified timepoints. In these zoomed-in views, the ordering of genes is fixed to the global TF heatmap for all three timepoints.
-
Targeted Dynamics: Users can query specific target genes or regulator TFs to generate dynamic, clustered heatmaps of individual regulatory axes. Interactive sliders allow users to define correlation or regression coefficient thresholds to filter transient noise.
-
Differential Topology: This component computes the change in Spearman correlation between any two TFs, at two user-selected timepoints. This allows the user to visually inspect TF-TF correlations with highest changes at the selected timepoints. The resulting topology is rendered dynamically using a constraint-based physics engine. Selecting a TF-TF edge in the graph displays the underlying metacell expression distributions of the TFs in the selected two timepoints, linking graph topology directly to transcriptomic evidence.
If you are interested in trying it out, Tangerine is on Github. To know more about how it works, read the preprint on Biorxiv.