How to change plotting order in plot_dimred, as in Seurat?

[erzakiev]

I want to change the plotting order in accordance to the expression level of a certain gene/set of genes:
cells expressing a certain gene at a higher level are plotted on top of those expressing it at a lower level.

Because I see that with the current plotting scheme the cells with low expression are often plotted above the cells with high expression of a certain gene.

In Seurat, for instance, the plotting order can be controlled with an argument called order

Seurat::FeaturePlot(x, "feature", cols = c("grey90", viridis::inferno(10000, begin = 0, end = 1, direction = -1)), reduction = "pca", dims = c(2,3))

When order=F

When ordering is to be done:

For the ggplot2, the basis of our plot_dimred, I found this post and tried both answers, none worked.

Reproducible example:

library(dyno)
smol <- readRDS(url('https://drive.google.com/u/0/uc?id=1ZjMOSiF2BULWIO4A5Ki58jGDfyMEXYAF&export=download'))
plot_dimred(smol, feature_oi = 'T_signature_gene')

[erzakiev]

i've since devised a function that rearranges BOTH trajectory and dimensional reduction, and that works:

ArrangeDataSetAccordingToFeature <- function(dataset, ordr){
  if(is.null(ordr)) return(dataset)
  if("dynwrap::with_expression" %in% class(dataset)){
    # if dataset is a dynwrap trajectory object
    dataset$cell_info <- dataset$cell_info[ordr,]
    dataset$grouping <- dataset$grouping[ordr]
    dataset$expression <- dataset$expression[ordr,]
    dataset$counts <- dataset$counts[ordr,]
    dataset$cell_ids <- dataset$cell_ids[ordr]
  } else {
    # dataset is dimensional reduction
    dataset <- dataset[ordr,]
  }
  return(dataset)
}

# the order of the feature in question
ord <- order(trajectory$expression[,feature], decreasing = F, na.last = F)

# changing the order for both the trajectory and the dimensional reduction
plot_dimred(ArrangeDataSetAccordingToFeature(trajectory, ord),
                    size_cells = 0.5,
                    dimred = ArrangeDataSetAccordingToFeature(dr, ord), 
                    feature_oi = feature,
                    plot_trajectory = F, 
                    hex_cells = F, 
                    color_cells = 'feature')