From 092d0adedd1fe5fdbcfb49310525d3183eeefd55 Mon Sep 17 00:00:00 2001
From: Michael Geuenich <m.geuenich@mail.utoronto.ca>
Date: Wed, 7 Dec 2022 21:54:07 -0500
Subject: [PATCH] viz doublet entropies

---
 .gitignore                                    |  2 +
 .../doublet-finder-entropies.R                | 45 +++++++++++++++++++
 2 files changed, 47 insertions(+)
 create mode 100644 pipeline/rem-cell-type-from-training/doublet-finder-entropies.R

diff --git a/.gitignore b/.gitignore
index a8c0258..66813eb 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,3 +4,5 @@ output/*
 .snakemake/*
 slurm*
 data/*
+.Rhistory
+
diff --git a/pipeline/rem-cell-type-from-training/doublet-finder-entropies.R b/pipeline/rem-cell-type-from-training/doublet-finder-entropies.R
new file mode 100644
index 0000000..fb0778c
--- /dev/null
+++ b/pipeline/rem-cell-type-from-training/doublet-finder-entropies.R
@@ -0,0 +1,45 @@
+suppressPackageStartupMessages({
+  library(tidyverse)
+})
+source("pipeline/whatsthatcell-helpers.R")
+
+doublets1 <- read_tsv("output/v6/results/doublet-id/scRNASeq-doublet-id-10x Chromium (v2) A.tsv")
+colnames(doublets1)[2:3] <- c("score", "classification")
+
+doublets2 <- read_tsv("output/v6/results/doublet-id/scRNASeq-doublet-id-10x Chromium (v2) B.tsv")
+colnames(doublets2)[2:3] <- c("score", "classification")
+
+doublets <- bind_rows(
+  doublets1,
+  doublets2
+)
+
+files <- list.files("output/v6/results/rem_cell_type/", full.names = TRUE)
+entropies <- lapply(files, read_tsv) |> 
+  bind_rows() |> 
+  mutate(al = case_when(grepl('AL_alg-multinom', params) ~ "multinom",
+                      grepl('AL_alg-rf', params) ~ 'rf'),
+       strat = case_when(grepl('strat-highest_entropy', params) ~ 'highest_entropy',
+                         grepl('strat-lowest_maxp', params) ~ 'lowest_maxp'),
+       init = case_when(grepl('init-random', params) ~ "random",
+                        grepl('init-ranking', params) ~ "ranking"),
+       ct = str_extract(params, 'rem_celltype.*'),
+       s = str_extract(params, '-seed-.*')) |>
+  mutate(ct = gsub("rem_celltype-", "", ct),
+         ct = gsub("-seed-[0-9]", "", ct),
+         s = gsub("-seed-", "", s)) |> 
+  select(-params)
+
+pdf(snakemake@output$pdf, height = 8, width = 20)
+  entropies |> 
+    left_join(select(doublets, -params), by = "cell_id") |> 
+    ggplot(aes(x = gt_cell_type, y = criterion_val, fill = classification)) +
+    geom_boxplot() +
+    labs(x = "Ground truth", y = "Entropy", "Doublet classification") +
+    facet_grid(num_missing_cells ~ ct) +
+    whatsthatcell_theme() +
+    theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5))
+dev.off()
+
+
+