* Support "multi-feature" analysis, e.g. parallel analysis of multiple
features (bins, peaks or gene) on the same object.
* New "Coverage" tab & functions generate_coverage_tracks() and
plot_coverage_BigWig() to generate cluster coverage tracks and interactively
visualise loci/genes of interest in the application.
* New inter- and intra-correlation violin plots to vizualise cell correlation
distribution between and within clusters.
* New normalization method : TF-IDF combined with systematic removal of PC1
strongly correlated with library size.
* Simple 'Copy Number Alteration' approximation & visualization using
'calculate_CNA' function for genetically re-arranged samples, provided one
or more control samples.
* New generate_analysis() & generate_report() functions to run a full-on
ChromSCape analysis and/or generate an HTML interactive report of an existing analysis.
* Supports 'custom' differential analysis to find differential loci between
a subset of samples and/or clusters.
* New pathway overlay on UMAP to visualize cumulative pathways signal
directly on cells.
* Now supports 'Fragment Files' input (e.g. from 10X cell ranger scATAC
pipeline), using a wrapper around 'Signac' package FeatureMatrix() function.
* New 'Contribution to PCA' plots showing most contributing features and
chromosome to PCA.
* Restructuration of the ChromSCape directory & faster reading/saving of
S4 objects using package 'qs'.
* RAM optimisation & faster pearson cell-to-cell correlations with 'coop'
package, and use of 'Rcpp' for as_dist() RAM-efficient distance calculation.
* Faster correlation filtering using multi-parallel processing.
* plot_reduced_dim now supports gene input to color cells by gene signal.
* All plots can now be saved in High Quality PDF files.
* Changed 'geneTSS' to 'genebody' with promoter extension to better reflect
the fact that mark spread in genebodies.
* Possibility to rename samples in the application.
* Downsampling of UMAPs & Heatmaps for fluider navigation.
* Changed 'total cell percent based' feature selection to manual selection of
top-covered features, as the previous was srongly dependent on the experiment size.
* Faster sparse SVD calculation.
* Faster differential analysis using pairWise Wilcoxon rank test from 'scran'
package.