update paper figs w paths/load libs

camlab-bioml · Jun 4, 2023 · e39fcf1 · e39fcf1
1 parent 6e8f7ef
commit e39fcf1
Show file tree

Hide file tree

Showing 10 changed files with 142 additions and 212 deletions.
diff --git a/pipeline/figures/AR-params.R b/pipeline/figures/AR-params.R
@@ -1,14 +1,15 @@
-suppressPackageStartupMessages(
+suppressPackageStartupMessages({
   library(tidyverse)
-)
+  library(patchwork)
+})
 source("pipeline/whatsthatcell-helpers.R")
 
 
-cytof_acc <- read_tsv("output/v7/results/overall-CyTOF-benchmarking-accuracies.tsv") |> 
+cytof_acc <- read_tsv(snakemake@input$cytof_acc) |> 
   mutate(cohort = "CyTOF")
-scrna_acc <- read_tsv("output/v7/results/overall-scRNASeq-benchmarking-accuracies.tsv") |> 
+scrna_acc <- read_tsv(snakemake@input$scrna_acc) |> 
   mutate(cohort = "scRNASeq")
-snrna_acc <- read_tsv("output/v7/results/overall-snRNASeq-benchmarking-accuracies.tsv") |> 
+snrna_acc <- read_tsv(snakemake@input$snrna_acc) |> 
   mutate(cohort = "snRNASeq")
 
 acc <- bind_rows(cytof_acc, scrna_acc, snrna_acc)
@@ -42,15 +43,15 @@ plot_knn_res <- function(acc, fill, cohort){
 }
 
 
-pdf("output/v7/paper-figures/supp-AR-res.pdf", height = 14, width = 9)
+pdf(snakemake@output$res, height = 14, width = 9)
   (plot_knn_res(cytof_acc, "res", "CyTOF") /
     plot_knn_res(scrna_acc, "res", "scRNASeq") /
     plot_knn_res(snrna_acc, "res", "snRNASeq")) +
     plot_layout(guides = "collect", heights = c(1, 2, 2))
 dev.off()
 
 
-pdf("output/v7/paper-figures/supp-AR-knn.pdf", height = 14, width = 9)
+pdf(snakemake@output$knn, height = 14, width = 9)
 (plot_knn_res(cytof_acc, "knn", "CyTOF") /
     plot_knn_res(scrna_acc, "knn", "scRNASeq") /
     plot_knn_res(snrna_acc, "knn", "snRNASeq")) +

diff --git a/pipeline/figures/cell-type-similarity.R b/pipeline/figures/cell-type-similarity.R
@@ -1,9 +1,9 @@
 library(tidyverse)
 library(ComplexHeatmap)
 
-cytof_sim <- read_tsv("output/v7/results/cell-type-similarity/similarity-CyTOF-seed-0.tsv")
-scrna_sim <- read_tsv("output/v7/results/cell-type-similarity/similarity-scRNASeq-seed-0.tsv")
-snrna_sim <- read_tsv("output/v7/results/cell-type-similarity/similarity-snRNASeq-seed-0.tsv")
+cytof_sim <- read_tsv(snakemake@input$cytof_sim)
+scrna_sim <- read_tsv(snakemake@input$scrna_sim)
+snrna_sim <- read_tsv(snakemake@input$snrna_sim)
 
 plot_dist_mat <- function(df){
   pivot_wider(df, names_from = "cell_type2", values_from = "cosine_similarity") |> 
@@ -14,14 +14,14 @@ plot_dist_mat <- function(df){
     Heatmap(name = "Cosine\nsimilarity")
 }
 
-pdf("output/v7/paper-figures/Supp-cell-type-sim-scRNASeq.pdf", height = 5, width = 5.5)
+pdf(snakemake@input$scrna, height = 5, width = 5.5)
   plot_dist_mat(scrna_sim)
 dev.off()
 
-pdf("output/v7/paper-figures/Supp-cell-type-sim-snRNASeq.pdf", height = 5, width = 5.5)
+pdf(snakemake@input$snrna, height = 5, width = 5.5)
   plot_dist_mat(snrna_sim)
 dev.off()
 
-pdf("output/v7/paper-figures/Supp-cell-type-sim-CyTOF.pdf", height = 5, width = 5.5)
+pdf(snakemake@input$cytof, height = 5, width = 5.5)
   plot_dist_mat(cytof_sim)
 dev.off()
diff --git a/pipeline/figures/compare-lr-rf-random-and-ranking.R b/pipeline/figures/compare-lr-rf-random-and-ranking.R
@@ -6,10 +6,11 @@ suppressPackageStartupMessages({
   library(magick)
   library(ggpubr)
 })
+devtools::load_all("/ggplot2")
 source("pipeline/whatsthatcell-helpers.R")
 
 ## Part A: Proportion of cells initially selected
-get_proportion_rep <- function(sce_path, selected_cells_path, cohort){
+get_proportion_rep <- function(sce_path, rand_files, rank_files, cohort){
   sce <- readRDS(sce_path)
 
   cell_types <- unique(sce$cell_type)
@@ -18,13 +19,6 @@ get_proportion_rep <- function(sce_path, selected_cells_path, cohort){
     cell_types <- unique(sce$CellType)
   }
 
-  rand_files <- list.files(selected_cells_path, 
-                           pattern = "random",
-                           full.names = TRUE)
-  rank_files <- list.files(selected_cells_path, 
-                           pattern = "ranking",
-                           full.names = TRUE)
-
   random <- lapply(rand_files, function(x){
     sel_cells <- read_tsv(x) |>
       filter(iteration == 0) |> 
@@ -48,14 +42,17 @@ get_proportion_rep <- function(sce_path, selected_cells_path, cohort){
 }
 
 props <- bind_rows(
-  get_proportion_rep("data/CyTOF/CyTOF-train-seed-0.rds", 
-                     "output/v8/data/CyTOF/Active-Learning_entropy/AL-batches-subset", 
+  get_proportion_rep(snakemake@input$cytof,
+                     snakemake@input$cytof_AL_rand,
+                     snakemake@input$cytof_AL_rank,
                      "CyTOF"),
-  get_proportion_rep("data/scRNASeq/scRNASeq-train-seed-0.rds", 
-                     "output/v8/data/scRNASeq/Active-Learning_entropy/AL-batches-subset", 
+  get_proportion_rep(snakemake@input$scrna,
+                     snakemake@input$scrna_AL_rand,
+                     snakemake@input$scrna_AL_rank,
                      "scRNASeq"),
-  get_proportion_rep("data/snRNASeq/snRNASeq-train-seed-0.rds", 
-                     "output/v8/data/snRNASeq/Active-Learning_entropy/AL-batches-subset", 
+  get_proportion_rep(snakemake@input$snrna,
+                     snakemake@input$snrna_AL_rand,
+                     snakemake@input$snrna_AL_rank,
                      "snRNASeq")
 )
 
@@ -72,11 +69,7 @@ prop <- props |>
 
 
 ## Part B: heatmaps
-f <- c("output/v8/results/overall-CyTOF-benchmarking-accuracies.tsv", 
-       "output/v8/results/overall-scRNASeq-benchmarking-accuracies.tsv", 
-       "output/v8/results/overall-snRNASeq-benchmarking-accuracies.tsv")
-
-acc <- lapply(f, function(x){
+acc <- lapply(snakemake@input$accs, function(x){
   df <- read_tsv(x) |> 
     mutate(cohort = case_when(grepl("CyTOF", basename(x)) ~ "CyTOF",
                               grepl("snRNASeq", basename(x)) ~ "snRNASeq",
@@ -149,7 +142,7 @@ ranking_vs_random <- create_heatmap(acc, "scRNASeq", "random_vs_ranked") +
 
 ranking <- draw(ranking_vs_random, column_title = "Selecting initial cells based on marker expression")
 
-pdf("output/v8/paper-figures/ranking-random-heatmap.pdf", height = 3.5, width = 7)
+pdf(snakemake@output$rank_rand_hm, height = 3.5, width = 7)
   ranking
 dev.off()
 
@@ -159,20 +152,20 @@ lr_vs_rf <- create_heatmap(acc, "scRNASeq", "lr_vs_rf") +
 
 lr_vs_rf <- draw(lr_vs_rf, column_title = "F1-score improvement by random forest compared to logistic regression")
 
-pdf("output/v8/paper-figures/lr-rf-heatmap.pdf", height = 3.5, width = 7)
+pdf(snakemake@output$lr_rf_hm, height = 3.5, width = 7)
   lr_vs_rf
 dev.off()
 
 ## Full main figure
-ranking <- image_read_pdf("output/v8/paper-figures/ranking-random-heatmap.pdf")
+ranking <- image_read_pdf(snakemake@output$rank_rand_hm)
 ranking <- ggplot() +
   background_image(ranking)
 
-lr_vs_rf <- image_read_pdf("output/v8/paper-figures/lr-rf-heatmap.pdf")
+lr_vs_rf <- image_read_pdf(snakemake@output$lr_rf_hm)
 lr_vs_rf <- ggplot() +
   background_image(lr_vs_rf)
 
-pdf("output/v8/paper-figures/lr-and-rf-heatmaps.pdf", height = 8.8, width = 7)
+pdf(snakemake@output$main_fig, height = 8.8, width = 7)
    (lr_vs_rf / wrap_elements(full = prop) / ranking) + plot_layout(heights = c(3, 1.2, 3)) + plot_annotation(tag_levels = "A")
 dev.off()
 
@@ -212,7 +205,7 @@ rf_vs_rf_ranking <- filter(acc, corrupted == 0) |>
   whatsthatcell_theme() +
   theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust =1))
 
-pdf("output/v8/paper-figures/rf-vs-lr-supplementary.pdf", height = 13, width = 12)
+pdf(snakemake@output$rf_lr_supp, height = 13, width = 12)
   rf_vs_rf_random / rf_vs_rf_ranking + plot_layout(guides = "collect") + 
     plot_annotation(tag_levels = "A")
 dev.off()
@@ -254,7 +247,7 @@ rand_vs_rank_rf <- filter(acc, corrupted == 0) |>
   whatsthatcell_theme() +
   theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust =1))
 
-pdf("output/v8/paper-figures/random-vs-ranking-supplementary.pdf", height = 13, width = 12)
+pdf(snakemake@output$rand_rank_supp, height = 13, width = 12)
   rand_vs_rank_lr / rand_vs_rank_rf + plot_layout(guides = "collect") + 
     plot_annotation(tag_levels = "A")
 dev.off()
diff --git a/pipeline/figures/figure1.R b/pipeline/figures/figure1.R
@@ -10,17 +10,18 @@ source("pipeline/whatsthatcell-helpers.R")
 
 
 ### [ FIGURE 1A - BENCHMARKING OVERVIEW ] #####
-schematic <- image_read("illustrator-figures/benchmarking-schematic.ai") |> 
-  image_ggplot()
+# had to remove because magick causes issues with docker
+# schematic <- image_read(snakemake@input$schematic) |> 
+#   image_ggplot()
 
 ### [ FIGURE 1B - DATASET COMPOSOTION ] #####
 set.seed(42)
 
-CyTOF <- readRDS("data/CyTOF/CyTOF-full.rds")
+CyTOF <- readRDS(snakemake@input$cytof)
 CyTOF <- runTSNE(CyTOF)
-scRNA <- readRDS("data/scRNASeq/scRNASeq-full.rds")
+scRNA <- readRDS(snakemake@input$scrna)
 scRNA <- runTSNE(scRNA)
-snRNA <- readRDS("data/snRNASeq/snRNASeq-full.rds")
+snRNA <- readRDS(snakemake@input$snrna)
 snRNA <- runTSNE(snRNA)
 
 
@@ -130,10 +131,10 @@ snrna_plot <- ((wrap_elements(full = snrna_tsne, ignore_tag = TRUE) & labs(title
                  snrna_bar) + 
   plot_layout(heights = c(3,1))
 
-pdf("output/v8/paper-figures/figure1.pdf", height = 12.5, width = 17.5)
+pdf(snakemake@output$fig1, height = 9.25, width = 17.5)
   wrap_elements((scrna_plot | snrna_plot | cytof_plot) + plot_layout(widths = c(1, 1.2, 1))) /
-    wrap_elements(schematic) +
-    plot_layout(heights = c(2, 0.7)) +
+    #wrap_elements(schematic) +
+    #plot_layout(heights = c(2, 0.7)) +
     plot_annotation(tag_levels = 'A') &
     theme(plot.tag = element_text(size = 22))
 dev.off()

diff --git a/pipeline/figures/figure3.R b/pipeline/figures/figure3.R
@@ -6,12 +6,11 @@ suppressPackageStartupMessages({
   library(scales)
   library(magick)
 })
+devtools::load_all("/ggplot2")
 source("pipeline/whatsthatcell-helpers.R")
 
 ### [ ACCURACIES ] ####
-f <- list.files("output/v8/results/", pattern = "overall", full.names = TRUE)
-
-acc <- lapply(f, function(x){
+acc <- lapply(snakemake@input$accs, function(x){
   df <- read_tsv(x) |> 
     mutate(cohort = case_when(grepl("CyTOF", basename(x)) ~ "CyTOF",
                               grepl("snRNASeq", basename(x)) ~ "snRNASeq",
@@ -34,13 +33,10 @@ acc <- lapply(f, function(x){
 
 sel_meth_cols <- sel_met_cols
 
-schematic <- image_read("illustrator-figures/AR-schematic.ai") |> 
-  image_ggplot()
-
 eval <- full_acc_plot_wrapper(acc, "rf", "ranking", "") &
   labs(fill = "Selection method")
 
-pdf("output/v8/paper-figures/figure3-overall-benchmark.pdf", height = 8.4, width = 9)
+pdf(snakemake@output$overall_fig, height = 8.4, width = 9)
   eval
 dev.off()
 
diff --git a/pipeline/figures/imbalance-plots.R b/pipeline/figures/imbalance-plots.R
@@ -1,9 +1,10 @@
 library(tidyverse)
 library(scales)
+library(patchwork)
 
-sc_acc <- read_tsv("output/v8/imbalance/acc/imbalance-acc-scRNASeq.tsv")
-sn_acc <- read_tsv("output/v8/imbalance/acc/imbalance-acc-snRNASeq.tsv")
-cy_acc <- read_tsv("output/v8/imbalance/acc/imbalance-acc-CyTOF.tsv")
+sc_acc <- read_tsv(snakemake@input$scrna_acc)
+sn_acc <- read_tsv(snakemake@input$snrna_acc)
+cy_acc <- read_tsv(snakemake@input$cytof_acc)
 
 source("pipeline/whatsthatcell-helpers.R")
 
@@ -79,7 +80,7 @@ plotting_order <- filter(comb_imb_score, modality == "snRNASeq" & al == "RF" & i
   arrange(mean_estimate) |> 
   pull(strat)
 
-pdf("output/v8/paper-figures/imbalance-main-figure.pdf", height = 3, width = 7)
+pdf(snakemake@output$main_fig, height = 3, width = 7)
   filter(comb_imb_score, modality == "snRNASeq" & al == "RF" & init == "ranking") |> 
     mutate(strat = factor(strat, levels = plotting_order)) |> 
     ggplot(aes(x = comp, y = diff, fill = strat)) +
@@ -95,7 +96,7 @@ pdf("output/v8/paper-figures/imbalance-main-figure.pdf", height = 3, width = 7)
 dev.off()
 
 
-pdf("output/v8/paper-figures/imbalance-supplementary.pdf", height = 18, width = 9)
+pdf(snakemake@output$sup_fig, height = 18, width = 9)
   (scrnaseq / snrnaseq / cytof) + 
     plot_annotation(tag_levels = "A") + 
   plot_layout(guides = "collect", heights = c(2, 2, 1))