name chunks in rmarkdown

saeyslab · Jan 26, 2024 · adf6fa6 · adf6fa6
1 parent d3df359
commit adf6fa6
Show file tree

Hide file tree

Showing 125 changed files with 554 additions and 513 deletions.
diff --git a/vignettes/circos.Rmd b/vignettes/circos.Rmd
@@ -1,5 +1,5 @@
 ---
-title: "Circos plot visualization to show active ligand-target links between interacting cells"
+title: "HNSCC case study + double circos visualization"
 author: "Robin Browaeys & Chananchida Sang-aram"
 date: "2023-07-20"
 output: rmarkdown::html_vignette
@@ -23,9 +23,9 @@ knitr::opts_chunk$set(
 )
 ```
 
-This vignette shows how to visualize the output of a NicheNet analysis in a circos plot. This is an additional demonstration to [Seurat Wrapper + Circos visualization](seurat_wrapper_circos.md). Here, we will use a normal matrix as input instead of a Seurat object. 
+This vignette shows how to visualize the output of a NicheNet analysis in a circos plot. This is an additional demonstration to [Seurat Wrapper + circos visualization](seurat_wrapper_circos.md). Here, we will use a normal matrix as input instead of a Seurat object. 
 
-We will use the same dataset as from [NicheNet's ligand activity analysis on a gene set of interest: predict active ligands and their target genes](ligand_activity_geneset.md) @puram_single-cell_2017. In contrast to the basic vignette, we will look at communication between multiple cell types. More specifically, we will predict which ligands expressed by both CAFs and endothelial cells can induce the p-EMT program in neighboring malignant cells.
+We will use the same dataset as from [NicheNet's ligand activity analysis on a gene set of interest](ligand_activity_geneset.md) @puram_single-cell_2017. In contrast to the basic vignette, we will look at communication between multiple cell types. More specifically, we will predict which ligands expressed by both CAFs and endothelial cells can induce the p-EMT program in neighboring malignant cells.
 
 ### Load packages
 
@@ -220,19 +220,19 @@ vis_circos_obj <- prepare_circos_visualization(circos_links,
 
 Render the circos plot where all links have the same transparency.  Here, only the widths of the blocks that indicate each target gene is proportional the ligand-target regulatory potential (~prior knowledge supporting the regulatory interaction). 
 
-```{r, fig.width=8, fig.height=8}
+```{r ligand-target-circos, fig.width=8, fig.height=8}
 draw_circos_plot(vis_circos_obj, transparency = FALSE,  args.circos.text = list(cex = 0.5)) 
 ```
 
 Render the circos plot where the degree of transparency determined by the regulatory potential value of a ligand-target interaction.
 
-```{r, fig.width=8, fig.height=8}
+```{r ligand-target-circos-transparent, fig.width=8, fig.height=8}
 draw_circos_plot(vis_circos_obj, transparency = TRUE,  args.circos.text = list(cex = 0.5)) 
 ```
 
 To create a legend for the circos plot, we can use the `ComplexHeatmap::Legend` function and creating a gTree object from it with `grid::grid.grabExpr`. As the circos plot is drawn on base R graphics (i.e., it is not a ggplot object), we will get the plot using `recordPlot()`.
 
-```{r, fig.show='hide'}
+```{r ligand-target-circos-unused, fig.show='hide'}
 par(bg = "transparent")
 
 # Default celltype order
@@ -245,24 +245,24 @@ circos_legend <- ComplexHeatmap::Legend(
   type = "point",
   grid_height = unit(3, "mm"),
   grid_width = unit(3, "mm"),
-  labels_gp = gpar(fontsize = 8)
+  labels_gp = grid::gpar(fontsize = 8)
   )
 
-circos_legend_grob <- grid::grid.grabExpr(draw(circos_legend))
+circos_legend_grob <- grid::grid.grabExpr(ComplexHeatmap::draw(circos_legend))
 
 draw_circos_plot(vis_circos_obj, transparency = TRUE, args.circos.text = list(cex = 0.5))
 p_circos_no_legend <- recordPlot()
 ```
 
 We can combine the circos plot and the legend using `cowplot::plot_grid`.
-```{r fig.width=8, fig.height=8}
+```{r ligand-target-circos-with-legend, fig.width=8, fig.height=8}
 cowplot::plot_grid(p_circos_no_legend, circos_legend_grob, rel_widths = c(1, 0.1))
 ```
 
 
 We can save this plot to an svg file.
 
-```{r}
+```{r, eval=FALSE}
 svg("ligand_target_circos.svg", width = 10, height = 10)
 cowplot::plot_grid(p_circos_no_legend, circos_legend_grob, rel_widths = c(1, 0.1))
 dev.off()
@@ -292,14 +292,14 @@ vis_circos_receptor_obj <- prepare_circos_visualization(lr_network_top_df,
 
 When drawing the plot, the argument `link.visible` = TRUE is also necessary for making all links visible, since no cutoff is used to filter out ligand-receptor interactions.
 
-```{r}
+```{r ligand-receptor-circos}
 draw_circos_plot(vis_circos_receptor_obj, transparency = FALSE,
                  link.visible = TRUE,  args.circos.text = list(cex = 0.8)) 
 ```
 
 Just as above, if `transparency = TRUE`, the degree of transparency is determined by the prior interaction weight of the ligand-receptor interaction.
 
-```{r}
+```{r ligand-receptor-circos-transparent}
 draw_circos_plot(vis_circos_receptor_obj, transparency = TRUE,
                  link.visible = TRUE,  args.circos.text = list(cex = 0.8)) 
 ```
@@ -388,7 +388,7 @@ vis_circos_obj_subset <- prepare_circos_visualization(circos_links_subset,
 
 Finally, we create the plot by adding an extra layer in `preAllocateTracks`, and we add a `for` loop at the end to draw the outer layer.
 
-```{r, fig.width=8, fig.height=8}
+```{r double-circos-method-1, fig.width=8, fig.height=8}
 circos.par(gap.degree = vis_circos_obj_subset$gaps)
 
 chordDiagram(vis_circos_obj_subset$links_circle,
@@ -500,7 +500,7 @@ vis_circos_obj2 <- prepare_circos_visualization(circos_links %>% select(-target_
 
 The general idea here is similar - adding an extra layer in `preAllocateTracks` and a `for` loop at the end to draw the outer layer - but the function allows for much cleaner code.
 
-```{r, fig.width=8, fig.height=8}
+```{r double-circos-method-2, fig.width=8, fig.height=8}
 circos.par(gap.degree = vis_circos_obj2$gaps)
 chordDiagram(vis_circos_obj2$links_circle,
              directional = 1,

diff --git a/vignettes/circos.md b/vignettes/circos.md
@@ -1,5 +1,4 @@
-Circos plot visualization to show active ligand-target links between
-interacting cells
+HNSCC case study + double circos visualization
 ================
 Robin Browaeys & Chananchida Sang-aram
 2023-07-20
@@ -10,12 +9,11 @@ rmarkdown::render("vignettes/circos.Rmd", output_format = "github_document")
 
 This vignette shows how to visualize the output of a NicheNet analysis
 in a circos plot. This is an additional demonstration to [Seurat
-Wrapper + Circos visualization](seurat_wrapper_circos.md). Here, we will
+Wrapper + circos visualization](seurat_wrapper_circos.md). Here, we will
 use a normal matrix as input instead of a Seurat object.
 
 We will use the same dataset as from [NicheNet’s ligand activity
-analysis on a gene set of interest: predict active ligands and their
-target genes](ligand_activity_geneset.md) Puram et al. (2017). In
+analysis on a gene set of interest](ligand_activity_geneset.md) Puram et al. (2017). In
 contrast to the basic vignette, we will look at communication between
 multiple cell types. More specifically, we will predict which ligands
 expressed by both CAFs and endothelial cells can induce the p-EMT
@@ -278,7 +276,7 @@ supporting the regulatory interaction).
 draw_circos_plot(vis_circos_obj, transparency = FALSE,  args.circos.text = list(cex = 0.5)) 
 ```
 
-![](circos_files/figure-gfm/unnamed-chunk-12-1.png)<!-- -->
+![](circos_files/figure-gfm/ligand-target-circos-1.png)<!-- -->
 
 Render the circos plot where the degree of transparency determined by
 the regulatory potential value of a ligand-target interaction.
@@ -287,7 +285,7 @@ the regulatory potential value of a ligand-target interaction.
 draw_circos_plot(vis_circos_obj, transparency = TRUE,  args.circos.text = list(cex = 0.5)) 
 ```
 
-![](circos_files/figure-gfm/unnamed-chunk-13-1.png)<!-- -->
+![](circos_files/figure-gfm/ligand-target-circos-transparent-1.png)<!-- -->
 
 To create a legend for the circos plot, we can use the
 `ComplexHeatmap::Legend` function and creating a gTree object from it
@@ -308,10 +306,10 @@ circos_legend <- ComplexHeatmap::Legend(
   type = "point",
   grid_height = unit(3, "mm"),
   grid_width = unit(3, "mm"),
-  labels_gp = gpar(fontsize = 8)
+  labels_gp = grid::gpar(fontsize = 8)
   )
 
-circos_legend_grob <- grid::grid.grabExpr(draw(circos_legend))
+circos_legend_grob <- grid::grid.grabExpr(ComplexHeatmap::draw(circos_legend))
 
 draw_circos_plot(vis_circos_obj, transparency = TRUE, args.circos.text = list(cex = 0.5))
 p_circos_no_legend <- recordPlot()
@@ -324,16 +322,14 @@ We can combine the circos plot and the legend using
 cowplot::plot_grid(p_circos_no_legend, circos_legend_grob, rel_widths = c(1, 0.1))
 ```
 
-![](circos_files/figure-gfm/unnamed-chunk-15-1.png)<!-- -->
+![](circos_files/figure-gfm/ligand-target-circos-with-legend-1.png)<!-- -->
 
 We can save this plot to an svg file.
 
 ``` r
 svg("ligand_target_circos.svg", width = 10, height = 10)
 cowplot::plot_grid(p_circos_no_legend, circos_legend_grob, rel_widths = c(1, 0.1))
 dev.off()
-## png 
-##   2
 ```
 
 ### Visualize ligand-receptor interactions of the prioritized ligands in a circos plot
@@ -369,7 +365,7 @@ draw_circos_plot(vis_circos_receptor_obj, transparency = FALSE,
                  link.visible = TRUE,  args.circos.text = list(cex = 0.8)) 
 ```
 
-![](circos_files/figure-gfm/unnamed-chunk-18-1.png)<!-- -->
+![](circos_files/figure-gfm/ligand-receptor-circos-1.png)<!-- -->
 
 Just as above, if `transparency = TRUE`, the degree of transparency is
 determined by the prior interaction weight of the ligand-receptor
@@ -380,7 +376,7 @@ draw_circos_plot(vis_circos_receptor_obj, transparency = TRUE,
                  link.visible = TRUE,  args.circos.text = list(cex = 0.8)) 
 ```
 
-![](circos_files/figure-gfm/unnamed-chunk-19-1.png)<!-- -->
+![](circos_files/figure-gfm/ligand-receptor-circos-transparent-1.png)<!-- -->
 
 ### Adding an outer track to the circos plot (ligand-receptor-target circos plot)
 
@@ -598,7 +594,7 @@ for (group in unique(targets_filtered$type)){
 }
 ```
 
-![](circos_files/figure-gfm/unnamed-chunk-22-1.png)<!-- -->
+![](circos_files/figure-gfm/double-circos-method-1-1.png)<!-- -->
 
 ``` r
 
@@ -691,7 +687,7 @@ for (target_gene_group in unique(target_gene_groups$target_type)){
 }
 ```
 
-![](circos_files/figure-gfm/unnamed-chunk-25-1.png)<!-- -->
+![](circos_files/figure-gfm/double-circos-method-2-1.png)<!-- -->
 
 ``` r
 

diff --git a/...s_files/figure-gfm/unnamed-chunk-22-1.png → ...s/figure-gfm/double-circos-method-1-1.png b/...s_files/figure-gfm/unnamed-chunk-22-1.png → ...s/figure-gfm/double-circos-method-1-1.png
diff --git a/...s_files/figure-gfm/unnamed-chunk-25-1.png → ...s/figure-gfm/double-circos-method-2-1.png b/...s_files/figure-gfm/unnamed-chunk-25-1.png → ...s/figure-gfm/double-circos-method-2-1.png
diff --git a/...s_files/figure-gfm/unnamed-chunk-18-1.png → ...s/figure-gfm/ligand-receptor-circos-1.png b/...s_files/figure-gfm/unnamed-chunk-18-1.png → ...s/figure-gfm/ligand-receptor-circos-1.png
diff --git a/...s_files/figure-gfm/unnamed-chunk-19-1.png → .../ligand-receptor-circos-transparent-1.png b/...s_files/figure-gfm/unnamed-chunk-19-1.png → .../ligand-receptor-circos-transparent-1.png
diff --git a/...s_files/figure-gfm/unnamed-chunk-12-1.png → ...les/figure-gfm/ligand-target-circos-1.png b/...s_files/figure-gfm/unnamed-chunk-12-1.png → ...les/figure-gfm/ligand-target-circos-1.png
diff --git a/...s_files/figure-gfm/unnamed-chunk-13-1.png → ...fm/ligand-target-circos-transparent-1.png b/...s_files/figure-gfm/unnamed-chunk-13-1.png → ...fm/ligand-target-circos-transparent-1.png
diff --git a/...s_files/figure-gfm/unnamed-chunk-14-1.png → ...ure-gfm/ligand-target-circos-unused-1.png b/...s_files/figure-gfm/unnamed-chunk-14-1.png → ...ure-gfm/ligand-target-circos-unused-1.png
diff --git a/...s_files/figure-gfm/unnamed-chunk-15-1.png → ...fm/ligand-target-circos-with-legend-1.png b/...s_files/figure-gfm/unnamed-chunk-15-1.png → ...fm/ligand-target-circos-with-legend-1.png
diff --git a/vignettes/images/nichenet_prior_model.png b/vignettes/images/nichenet_prior_model.png
diff --git a/vignettes/ligand_activity_geneset.Rmd b/vignettes/ligand_activity_geneset.Rmd
@@ -185,7 +185,7 @@ We will use the top 30 ligands to predict active target genes and construct an a
 
 Active target genes are defined as genes in the gene set of interest that have the highest regulatory potential for each top-ranked ligand. These top targets of each ligand are based on the prior model. Specifically, the function get_weighted_ligand_target_links will return genes that are in the gene set of interest and are the top `n` targets of a ligand (default: `n = 200`).
 
-```{r, fig.width=8, fig.height=6}
+```{r ligand-target-heatmap, fig.width=8, fig.height=6}
 active_ligand_target_links_df <- best_upstream_ligands %>%
   lapply(get_weighted_ligand_target_links,
          geneset = geneset_oi,
@@ -212,7 +212,7 @@ p_ligand_target_network
 
 We can also look at which receptors of the receiver cell population (malignant cells) can potentially bind to the prioritized ligands from the sender cell population (CAFs).
 
-```{r, fig.width=9, fig.height=6}
+```{r ligand-receptor-heatmap, fig.width=9, fig.height=6}
 
 ligand_receptor_links_df <- get_weighted_ligand_receptor_links(
   best_upstream_ligands, expressed_receptors,
@@ -247,7 +247,7 @@ vis_ligand_aupr <- ligand_activities %>% filter(test_ligand %in% best_upstream_l
   column_to_rownames("test_ligand") %>% select(aupr_corrected) %>% arrange(aupr_corrected) %>% as.matrix(ncol = 1)
 ```
 
-```{r, fig.width=5, fig.height=6}
+```{r ligand-activity-heatmap, fig.width=5, fig.height=6}
 p_ligand_aupr <- make_heatmap_ggplot(vis_ligand_aupr,
                                      "Prioritized CAF-ligands", "Ligand activity",
                                      color = "darkorange", legend_title = "AUPR") + 
@@ -280,7 +280,7 @@ vis_ligand_tumor_expression <- aggregated_expression_matrix_CAF[rev(best_upstrea
 
 ```
 
-```{r, fig.width=9, fig.height=6}
+```{r ligand-expression-heatmap, fig.width=9, fig.height=6}
 color <- colorRampPalette(rev(brewer.pal(n = 7, name ="RdYlBu")))(100)
 p_ligand_tumor_expression <- make_heatmap_ggplot(vis_ligand_tumor_expression,
                                                  "Prioritized CAF-ligands", "Tumor",
@@ -310,7 +310,7 @@ vis_target_tumor_expression_scaled <- aggregated_expression_matrix_target %>% t(
   .[order_tumors, order_targets]
 ```
 
-```{r, fig.width=9, fig.height=6}
+```{r target-expression-heatmap, fig.width=9, fig.height=6}
 p_target_tumor_scaled_expression <- make_threecolor_heatmap_ggplot(vis_target_tumor_expression_scaled,
                                                                    "Tumor", "Target",
                                                                    low_color = color[1], mid_color = color[50], mid = 0.5,
@@ -322,7 +322,7 @@ p_target_tumor_scaled_expression
 
 #### Combine the different heatmaps in one overview figure
 
-```{r, fig.width=15, fig.height=9}
+```{r summary-vis, fig.width=15, fig.height=9}
 
 figures_without_legend = plot_grid(
   p_ligand_aupr + theme(legend.position = "none"),