add sc_geom_annot to add the annotation layer for external data

[xiangpin]

This pr provides two methods like the methods of ggtree linking to external data

• using data of sc_geom_annot (recommend)
• using %<<+% method to add the external data to the plot data.

Annotation for single feature

{library(SingleCellExperiment)
library(SpatialExperiment)
library(SVP)
library(ggsc)
library(ggplot2)} |> suppressPackageStartupMessages()

data(hpda_spe_cell_dec)
hpda_spe_cell_dec

# identifing the spatial variable function
hpda_spe_cell_dec <- hpda_spe_cell_dec |>
                     runDetectSVG(assay.type="affi.score", method='moran', weight.method='voronoi', action='add')

# extrating the result (action = 'add')
svg.res <- hpda_spe_cell_dec |> svDf()
svg.res |> dplyr::arrange(rank) |> head()


# identifing the spatial functional domain 
lisa.res <- hpda_spe_cell_dec |>
            runLISA(assay.type = 'affi.score', features='Cancer clone A', method='localG', weight.method='voronoi')

# this is the result of local spatial stastics using localG.
lisa.res
lisa.res[["Cancer clone A"]] |> head()
dt <- lisa.res[['Cancer clone A']] |> tibble::rownames_to_column(var='.BarcodeID')

p1 <- sc_spatial(hpda_spe_cell_dec, features='Cancer clone A', mapping=aes(x=x, y=y), pointsize=8)
p1

The first method:

the dt (external data) which must contain .Barcode and features or (only .Barcode for single feature), will be added to the f1$data using left_join internally.

p2 <- p1 +
      sc_geom_annot(
        data = dt,
        mapping = aes(bg_colour=cluster.test, subset=cluster.test %in% c("High", "Low")),
        pointsize = 7,
        bg_line_width = .14,
        gap_line_width = .12
      ) +
      scale_bg_color_manual(values=c('black', 'grey'))

The second method:

the dt (external data) will be added to the f1$data using %<<+% (also using left_join), but change the plot data. (not recommend)

p1 <- p1 %<<+% dt

p3 <- p1 +
      sc_geom_annot(
        mapping = aes(bg_colour=cluster.test, subset=cluster.test %in% c("High", "Low")),
        pointsize = 7,
        bg_line_width = .14,
        gap_line_width = .12
      ) +
      scale_bg_color_manual(values = c("black", "grey"))
p3

Annotation for multiple features

we can also perform the local spatial statistics to identify the functional spatial domain using runLISA of SVP

lisa.res2 <- runLISA(hpda_spe_cell_dec,
                     features=rownames(hpda_spe_cell_dec)[seq(6)],
                     assay.type='affi.score')
lisa.res2
df <- lisa.res2 |> lapply(function(x)x|>tibble::rownames_to_column(var='.BarcodeID')) |> dplyr::bind_rows(.id='features')
df |> head()
f1 <- sc_spatial(hpda_spe_cell_dec, features = names(lisa.res2), mapping=aes(x=x, y=y), pointsize = 9, ncol=3)
f1

The first method:

using sc_geom_annot(data = df, ...)

f2 <- f1 +
      sc_geom_annot(
        data = df,
        mapping = aes(bg_colour = cluster.test, subset = cluster.test %in% c("High", "Low")),
        pointsize = 7,
        bg_line_width = .15,
        gap_line_width = .12
      ) +
      scale_bg_color_manual(values = c("black", "grey"))
f2

The second method:

using %<<+% and sc_geom_annot(...)

f1 <- f1 %<<+% df

f3 <- f1 +
    sc_geom_annot(
        mapping = aes(bg_colour = cluster.test, subset = cluster.test %in% c("High", "Low")),
        pointsize = 7,
        bg_line_width = .15,
        gap_line_width = .12
    ) +
    scale_bg_color_manual(values = c("black", "grey"))
f3

When the common.legend=FALSE

the fig1 is a patchwork object when common.legend=FALSE.

fig1 <- sc_spatial(hpda_spe_cell_dec,
                   features = names(lisa.res2),
                   mapping = aes(x=x, y=y),
                   pointsize = 9,
                   common.legend=FALSE,
                   ncol = 3
        )

the first method:

the annotation layer should use & to add the original layer instead of +.

fig2 <- fig1 &
        sc_geom_annot(
          data = df,
          mapping = aes(bg_colour = cluster.test, subset = cluster.test %in% c("High", "Low")),
          pointsize = 7,
          bg_line_width = .14,
          gap_line_width = .12
        ) &
        scale_bg_color_manual(values = c('black', 'grey'))

the second method:

fig1 <- fig1 %<<+% df
fig3 <- fig1 &
      sc_geom_annot(
        mapping = aes(bg_colour = cluster.test, subset = cluster.test %in% c("High", "Low")),
        pointsize = 7,
        bg_line_width = .14,
        gap_line_width = .12
      )
     
fig3

[GuangchuangYu]

这个%<<+%能和ggtree::%<+%统一？

可以统一的话，我们可以放到ggfun包里。

[xiangpin]

我看下，应该也可以用S3方法写成统一的

[xiangpin]

已经用S3方法统一到ggfun中

[GuangchuangYu]

那么ggtree和ggsc就做相应的修改吧。

[xiangpin]

已经做好相应的修改了，现在统一成 %<+% , 所以上面ggsc 中有用到%<<+%改成%<+%即可。然后再DESCRIPTION里已经声明ggfun (>0.1.4).