Major version bump!
- Added support for Phylogenies as latent spaces in core VISION and
integrated with VISION api
- Re-engineered UI into Signature Autocorrelation and Hotspot
views
- Integrated Hotspot into
VISION analysis and report UI.
- Deprecated support for trajectories and LC Annotator.
- New color-scheme for gene expression values
Added parameter sig_gene_threshold with changed
default behavior
- Before it was a guideline to filter lowly expressed genes before
running VISION
- Now, by default, genes expressed in fewer than 0.1% of cells will be
filtered automatically
Bug Fixes:
- Better colors in output when more than 10 categories
- Errors with certain output object accessors
- Selections not saving when loading results server
- Crashes when running on more than 1200 signatures
Lots of changes for this version.
However the object structure has been refactored. This means that any
old saved RDS objects will not work with the viewResults function once
the code has been upgraded. To resolve this, either re-create the
objects or use the latest v1.x release of VISION. Incompatible changes
like this will be avoided in the future, but a refactor was long
needed.
- Added support for surface protein data (e.g., CITE-seq)
- This data is entered separately using the
proteinData
argument
- Autocorrelation scores will be computed on protein data
- Differential expression tests will also test protein data
- Can view protein vs. protein (FACS-style) in output report
- Improved handling of sparse data
- When inputing sparse expression data, the processing pipeline will
no longer expand this data to dense at any stage
- Additional improvements for performance and memory usage
- New exported pipeline functions for custom workflows
- UI Improvements
- Can see multiple, different views simultaneously in the output (by
unchecking the ‘Update All?’ changes only affect the selected plot
- Y-axis for feature histograms can be log-scaled
- Filtering (FDR) and Export for DE results table
- Caching of DE results to avoid re-running
- Additional DE options (subsampling, gene filtering) for reduced run
times
- Improved signature-gene heatmap
Many thanks to @Yanay1 for improvements to the DE results and
heatmap!
- Bugfixes
- Fixed issue with selection on high-dpi displays
- Better Encoding of URI Components
- Removed gzmem as a suggested dependency
- Various style fixes
- Added support for Seurat 3.X objects
- Added differential expression testing to the output report
- Added gene-signature importance calculation
- More flexible projection plotting
- Added UMAP as a projection method
- Bugfixes related to caching in the output report
- Change default
cellsPerPartition value to 10
- Change default
filterThreshold parameter value to 5% in
applyMicroClustering (for consistency with Vision constructor)
- Added Layout Options to the output report
- Selected cells can now be saved/loaded
- Addition of “get*“-style accessor methods to the VISION object
- Added an interface to Seurat objects
- Added convenience methods for working with outputs from 10x
Genomic’s CellRanger
-
Note: Some changes were made to the Vision object
structure and how results are stored. If updating to 1.0.0, older Vision
objects may need to be re-created and re-run before using
viewResults