2.0.0

SCpubr v2.0.0

This major update focus on a complete re-implementation of all heatmap-based functions into ggplot2 instead of ComplexHeatmap. This will lead to many of the existing code to break. The trade-off between the difficulty of debug, expand and maintain the existing heatmap-based functions with regards to the capabilities ComplexHeatmap offers with regards to ggplot2 was not worthy.

All heatmap-specific parameters have been replaced with the overarching parameters that are used across functions. This decision was taking after a lot of thought, but ultimately, having all plots rely on ggplot2 makes it way more compatible to work with them together, to debug, and to further implement new ideas.

Many (except a few selected cases) of the functions that returned list of different plots have been modified to return a single (and most important/relevant) plot and the option to return the Seurat object with the data generated added to it has been implemented so that the user can still generate plots with it. This goes in line with the fact that having so many interconnected functions made it very difficult to expand on them, if needed, as the downstream effects will cascade to other functions as well.

Parameter renaming

• Changed viridis_color_map to viridis.palette.
• Changed viridis_direction to viridis.direction.
• Changed sequential_direction to sequential.direction.
• Changed rotate_x_axis_labels to axis.text.x.angle.
• Changed rotate_strip_text to strip.text.angle.

New functions (available on the development build for extended texting)

• SCpubr::do_MetadataPlot() to generate metadata heatmaps with ease both from Seurat object or from a data frame. Will be first released as part of the development version and then released in CRAN as part of future updates. The idea is to gather feedback from users before officially releasing it.
• SCpubr::do_SCExpressionHeatmap() to generate heatmaps of expression of genes across all cells in the dataset. Will be first released as part of the development version and then released in CRAN as part of future updates. The idea is to gather feedback from users before officially releasing it.
• SCpubr::do_SCEnrichementHeatmap() to generate heatmaps of enrichment of genes across all cells in the dataset. Will be first released as part of the development version and then released in CRAN as part of future updates. The idea is to gather feedback from users before officially releasing it.
• SCpubr::do_AffinityAnalysisPlot() to assess the affinity of gene sets to subset of cells in the Seurat objects using the weighted means algorithms from DecoupleR. Will be first released as part of the development version and then released in CRAN as part of future updates. The idea is to gather feedback from users before officially releasing it.
• SCpubr::do_LoadingsPlot() to generate a summary heatmap of the PCA loadings (top and bottom scored genes for each PC) together with a expression heatmap of the same genes. Will be first released as part of the development version and then released in CRAN as part of future updates. The idea is to gather feedback from users before officially releasing it.
• SCpubr::do_DiffusionMapPlot() to analyze the output of a diffusion map analysis on the context of enrichment in gene sets used for the generation of the diffusion map. Will be first released as part of the development version and then released in CRAN as part of future updates. The idea is to gather feedback from users before officially releasing it.
• SCpubr::check_dependencies() to generate a per-function summary of the needed packages to run the function. The report has been enhanced with cli package and now clearly illustrates what is missing to run the function.

Removed functions

• SCpubr::do_SankeyPlot() has been removed and replaced by SCpubr::do_AlluvialPlot(), which is present in the official CRAN versions.
• SCpubr::do_PseudotimePlot() has been removed indefinitely until a better, revamped, state-of-the-art version is generated.
• SCpubr::do_AzimuthAnalysisPlot() has been removed as the output can be accomplished by a combination of the current functions in SCpubr. A vignette will be added to reproduce the same analysis.

General

• Now when using min.cutoff or max.cutoff, the legend will show that the min/max value is higher/lower than the one provided, if such value appeared originally in the legend breaks. This potentially interacts with enforce_symmetry.
• Added number.breaks parameter to control the number of breaks in the legend of ggplot2-based plots. It will not always work, as the function will try to fit the breaks accordingly. But still, will give some range of freedom to the user.
• Removed colorsteps from legend.type parameters as it was prone to generate unintended bugs.
• Changed default values from min.cutoff and max.cutoff from NULL to NA.
• Implemented diverging.palette parameter in all plots that have a symmetrical color scale to help selecting other possible color scales for the plot.
• Implemented sequential.palette parameter in all plots that have a continuous, non-symmetrical color scale to help selecting other possible color scales for the plot, in the case the user does not want to use viridis color scales.
• Renamed SCpubr::state_dependencies() to SCpubr::check_dependencies().
• Renewed printed messages at startup and while running functions using cli package.
• Added the complete control of the font style of plot titles, subtitles, captions, axis titles, axis text, legend titles and legend text. For this, the following parameters have been added to all ggplot2-based functions:
  • plot.title.face: To control the style of the title.
  • plot.subtitle.face: To control the style of the subtitle.
  • plot.caption.face: To control the style of the caption.
  • axis.title.face: To control the style of the axis title.
  • axis.text.face: To control the style of the axis text.
  • legend.title.face: To control the style of the legend title.
  • legend.text.face: To control the style of the legend text.
• Changed default font style for legend text from bold to plain.
• Changed default font style for axis text from bold to plain.
• When using plot.axes = TRUE parameter in SCpubr::do_DimPlot(), SCpubr::do_FeaturePlot() and SCpubr::do_NebulosaPlot(), now the entirety of the X and Y axis is removed, titles included.
• Remove plot margin padding in SCpubr::do_DimPlot(), SCpubr::do_FeaturePlot() and SCpubr::do_NebulosaPlot().

SCpubr::do_AlluvialPlot

• Added sequential.palette and sequential.direction parameters.

SCpubr::do_BarPlot

• Added facet.by parameter to extra group the bars by a third metadata variable.
• Added facet.by.direction parameter to decide in which direction the facets are drawn.
• Added order.by to reorder the bars when using position = fill based on a value in group.by.
• Limited the possible interactions from group.by, split.by and order.by to those that make sense to plot. For instance, a bar plot using group.by and position = fill but not using split.by resulted in bars of equal lenght with only one value per group of proportion 1.
• Set default value of plot.grid to FALSE.
• Added parameter add.n to display the total count on top when position = fill.
• Added parameter add.n.face to control the appearance of the text displayed.
• Added parameter add.n.expand to control the range of values in the Y axis. This has to be minimum 0 and maximum at least 1. This is set in order to tweak the limits so that the labels fit when flip = TRUE.

SCpubr::do_BeeSwarmPlot

• Added order parameter to reorder the groups based on the median rank.

SCpubr::do_BoxPlot

• Changed the reordering of boxplots based on the median rather than the mean.
• Added na.rm to geom_boxplot to avoid unnecessary warnings when introducing NAs as part of the data.
• Fixed a bug in which order would not work if NAs are in the data.
• Changed default value of boxplot.linewidth from 1 to 0.5.
• Fixed a bug in which when using a combination of group.by and split.by, the package would check that the colors provided to colors.use need to match the values in group.by and not split.by.

SCpubr::do_CorrelationPlot

• Added parameter to fix a bug in which viridis scales did not apply due to the lack of the parameter.
• Added min.cutoff and max.cutoff parameter to add cutoffs to the scales.
• Added mode = "jaccard" to compute a correlation matrix of a list of gene sets based on jaccard similarity.
• Added use_viridis, sequential.palette and sequential_direction and diverging.palette to control color palettes.
• Added cluster parameter to toggle on/off the clustering of the rows and columns in the heatmap.
• Added remove.diagonal parameter to toggle on/off the conversion of the diagonal in the correlation matrix to NA.
• Fixed several issues with setting cutoffs for the color scale using min.cutoff and max.cutoff.
• Fixed an issue where number.breaks will not work in mode = "jaccard".

SCpubr::do_CopyNumberVariantPlot()

• Removed the option to compute Feature and Geyser plots.
• Instead, a new paramerter return_object has been added to return the Seurat object with a new assay containing the CNV scores per cell on the data slot of the CNV_scores assay.
• The main output visualization is now a heatmap with the averaged scores by chromosome and groups and also by chromosome arms.

SCpubr::do_DimPlot

• Modified underlying code to correctly display borders around cells when cells.highlight or idents.hightlight or idents.keep are used. Also removed the "Not selected" item from the legend when doing so, as it was redundant.
• Fixed a bug in which multiple legend would appear when using a combination of group.by and split.by, given that the individual UMAPs would not have the same number of entities to plot and color.

SCpubr::do_DotPlot

• Added scale parameter to allow for the data to be scaled or not scaled.
• Removed split.by parameter in favor or the higher consistency and proper functionality accross parameters. Will probably come in the future, implemented outside of the umbrella of Seurat.
• Renamed parameter cluster.idents to cluster.
• Removed the limitation of flip when features was a list of genes. Now any combination of flip and features is possible.

SCpubr::do_EnrichmentHeatmap

• Removed options to plot FeaturePlots, GeyserPlots, ViolinPlots, etc. - together with its related parameters. For the sake of simplicity in the function and its use, the user can get the Seurat object back with return_object = TRUE and plot the enrichment scores separately, that are stored as a new Assay.
• Removed return_matrix parameter as the scores can now be retrieved from the Seurat object as an assay.
• Enforcing the use of named lists as input for the function.
• Added cluster parameter to allow for clustering of rows and columns.
• Added groups.order to allow for specifically arrange the groups defined by group.by with a given order.
• Added features.order to allow for specifically arrange the gene sets defined by input_gene_list.

SCpubr::do_ExpressionHeatmap

• Added cluster parameter to allow for clustering of rows and columns.
• Added groups.order to allow for specifically arrange the groups defined by group.by with a given order.
• Added features.order to allow for specifically arrange the features defined by features.

SCpubr::do_FeaturePlot

• Modified underlying code to show a border around selected cells when using split.by, cells.hightlight and idents.highlight.
• Added parameter border.density to reduce the amount of extra cells drawn on the background to generate the borders. This will be a number between 0 and 1 corresponding to the quantile of the distribution of density of the points in the scatterplot drawn in the background. The lower the value, the harder it will be to keep a border around all cells, while it will significantly reduce the overall weight of the plot object.
• Added parameter group.by, that allows to plot a big dot in the center of each group designated by group.by and thus allowing to locate easily where each identity is in the FeaturePlot. Also, plots a legend matching the color of the dots. This can be tweaked with additional parameters such as:
• group.by.show.dots to controlw hether these dots are plotted or not (to allow only plotting colored borders around cells - see below).
• group.by.dot.size to control the size of the introduced dots.
• group.by.cell_border to plot another contour-like border which also displays the color coding of the clusters designated by group.by, to signal the reach of each cluster. However, this basically signals the cluster the cells in the periphery of the cell blobs belong to. Take that into account.
• group.by.cell_borders.alpha controls the alpha of the new cell borders.
• group.by.legend controls the legend title of the new legend.
• Renamed split.by.idents to idents.keep to better synergize with the parameter in SCpubr::do_DimPlot. Only works when split.by is used.

SCpubr::do_FunctionalAnnotationPlot

• Removed the tree plots as they proved to behave inconsistently across datasets and the quality of visualizations were compromised.
• Removed the option to plot the bar plots and dot plots in the sake of a more simplified, streamlined plot generation.
• The option to return the result matrix using return_matrix is added, so that the user can use it to compute further analysis or visualizations.

SCpubr::do_FunctionalAnnotationPlot

• Renamed order_by_mean to order.
• Ordering using order = TRUE now is done based on the median instead of the mean.

SCpubr::do_LigandReceptorPlot()

• Modified the accepted input so that only the result of liana::liana_aggregate() is taken into account.
• Removed arrange_interactions_by as now the function only accepts the output of liana::liana_aggregate().
• Added a sort.by parameter instead to select how the output of liana::liana_aggregate() should be ordered prior the subset by top_interactions. Five modes are available:
  • A: Orders the output by specificity.
  • B: Orders the output by magnitude.
  • C: Orders the output by specificity then magnitude. This prioritizes the specificity column.
  • D: Orders the output by magnitude then specificity. This prioritizes the magnitude column.
  • E: Orders the output by specificity and magnitude providing equal weights to both columns.
• Removed flip parameter as the output was prone to errors.
• Removed parameter compute_ChordDiagrams and added return_interactions. This parameter returns two tibbles that can be used alongside SCpubr::do_ChordDiagramPlot to plot the diagrams.
• Now the filtering applied by using keep_source and keep_target takes place before subsetting for the top N interactions defined by top_interactions. This ensures that, if the user wants to focus on a given interaction subset, we retrieve the most important interactions for the subset.
• Added magnitude and specificity columns to allow the user to choose which variables to use for plotting.
• Added sorting.type.magnitude and sorting.type.specificity to allow the user to choose how the columns are sorted prior plotting.
• Added invert_magnitude and invert_specificity to allow the user to choose how the data is displayed for columns that tend to 0. Inverting performs a -log10 transformation on the column.
• Added a verbose parameter and set it to TRUE by default to inform the user of the arrangements taking place in the output of liana::liana_aggregate() prior plotting.

SCpubr::do_PathwayActivityPlot()

• Removed the option to plot geyser and feature plots to simplify the use (and computational time) of the function.
• Introduced return_object parameter that returns the Seurat object with the new assay to use for other plotting purposes (such as Geyser and Feature plots).
• Removed options to plot FeaturePlots, GeyserPlots - together with its related parameters. For the sake of simplicity in the function and its use, the user can get the Seurat object back with return_object = TRUE and plot the scores separately.
• Added slot parameter to decide whether to plot scale data or not.
• Fixed bug in which after setting enforce_symmetry = FALSE the color palette used was diverging.palette instead.

SCpubr::do_TFActivityPlot()

• Removed the option to plot geyser and feature plots to simplify the use (and computational time) of the function.
• Introduced return_object parameter that returns the Seurat object with the new assay to use for other plotting purposes (such as Geyser and Feature plots).
• Removed options to plot FeaturePlots, GeyserPlots - together with its related parameters. For the sake of simplicity in the function and its use, the user can get the Seurat object back with return_object = TRUE and plot the scores separately.
• Added slot parameter to decide whether to plot scale data or not.
• Fixed bug in which after setting enforce_symmetry = FALSE the color palette used was diverging.palette instead.