Merge pull request #11 from stemangiola/edit_from_local

updates in the local computer

Author: stemangiola

dev/.gitignore

@@ -6,3 +6,4 @@ slurm.wrapper
 ERROR-1.output
 core.26183
 curated_annotation.rds
+metadata.sqlite

dev/annotation_harmonise.R

@@ -15,6 +15,9 @@ library(tidyseurat)
 library(celldex)
 library(SingleR)
 library(glmGamPoi)
+library(stringr)
+library(purrr)
+
 # source("utility.R")
 
 clean_cell_types = function(.x){
@@ -557,24 +560,66 @@ curated_annotation =
 		cell_type_harmonised
 	)) |>
 	select(-.cell_combined) |>
-	select(-cell_type_Monaco, -score)
-
+	select(-cell_type_Monaco, -score) |> 
+  
+  # Replace NA
+  mutate(cell_type_harmonised = if_else(is.na(cell_type_harmonised), "immune_unclassified", cell_type_harmonised)) |>
+  
+  # Add non immune
+  select(-cell_type) |> 
+  full_join(
+    get_metadata("dev/metadata.SQLite") |> 
+      select(.cell, .sample) |> 
+      as_tibble()
+  ) |> 
+  mutate(cell_type_harmonised = if_else(is.na(cell_type_harmonised), "non_immune", cell_type_harmonised))
+
+# Save
 job::job({
 	curated_annotation |>
+    mutate(across(contains("cell_"), as.factor)) |> 
 	saveRDS("dev/curated_annotation.rds")
 })
 
 
 cell_metadata_with_harmonised_annotation =
 	curated_annotation |>
+  mutate(.cell = .cell |> str_remove(.sample) |> str_remove("_$")) |> 
 	left_join(
-		get_metadata() |>
-			select(.cell, .sample, file_id, file_id_db, tissue) |>
-			as_tibble()
-	)
-# xx = x |>
-# 	filter(cell_type_harmonised == "monocytes") |>
-# 	get_SingleCellExperiment()
+		get_metadata("dev/metadata.SQLite") |>
+			select(.cell, .sample, file_id, file_id_db, tissue, disease, is_primary_data.x, is_primary_data.y, name) |>
+		  left_join(read_csv("dev/tissue_label_curated.csv"), copy=TRUE) |> 
+		  as_tibble(),
+		by=c(".cell", ".sample")
+	) |> 
+  distinct()  |> 
+  
+  # Drop secondary data often cell type subsets
+  filter(is_primary_data.x==TRUE)
+
+# Filter samples that do not have immune
+cell_metadata_with_harmonised_annotation = 
+  cell_metadata_with_harmonised_annotation |> 
+  nest(data = -c(.sample, tissue_harmonised)) |> 
+  filter(map_int(data, ~ .x |> filter(cell_type_harmonised != "non_immune") |> nrow()) > 0) |> 
+  unnest(data)
+
+# Tissue with no immune
+cell_metadata_with_harmonised_annotation = 
+  cell_metadata_with_harmonised_annotation |> 
+  filter(disease == "normal") |> 
+
+  # Filter tissues
+  nest(data = -c(cell_type_harmonised, tissue_harmonised)) |> 
+  add_count(tissue_harmonised, name = "n_cell_type_in_tissue") |> 
+  filter(n_cell_type_in_tissue>=18) |> 
+  add_count(cell_type_harmonised, name = "n_tissue_in_cell_type") |> 
+  filter(n_tissue_in_cell_type>=26) |> 
+  unnest(data)
+
+
+cell_metadata_with_harmonised_annotation |> 
+  saveRDS("dev/cell_metadata_with_harmonised_annotation.rds")
 
 
 
@@ -590,9 +635,24 @@ cell_metadata_with_harmonised_annotation |> anti_join(annotated_samples)  |>  di
 # Histo of annotation
 cell_metadata_with_harmonised_annotation |> filter(!is.na(cell_type_harmonised)) |>  distinct( cell_type_harmonised, .sample) |> count(.sample) |> pull(n) |> hist(breaks=30)
 
+# NEEDED CHANGES - ANIMAL CELLS AND HEMATOPOIETIC SHOULD BE EVALUATED AS IMMUNE FOR REANNNOTATION?
+
 # Tissue with no immune
-cell_metadata_with_harmonised_annotation |> filter(!is.na(cell_type_harmonised)) |>  distinct( cell_type_harmonised, tissue_harmonised) |> count(cell_type_harmonised) |> arrange(n) |> print(n=99)
 
-cell_metadata_with_harmonised_annotation |> filter(!is.na(cell_type_harmonised)) |>  distinct( cell_type_harmonised, tissue_harmonised) |> count(tissue_harmonised) |> arrange(n) |> print(n=99)
+temp_tissue = 
+  cell_metadata_with_harmonised_annotation |> 
+  filter(cell_type_harmonised=="pdc") |> 
+  pull(tissue_harmonised)  
+  
+cell_metadata_with_harmonised_annotation |> 
+  filter(!tissue_harmonised %in% temp_tissue) |> 
+  distinct(tissue_harmonised) |> 
+  
+
+cell_metadata_with_harmonised_annotation |> 
+  filter(disease == "normal") |> 
+  filter(!is.na(cell_type_harmonised)) |>  
+  distinct( cell_type_harmonised, tissue_harmonised) |>
+  count(tissue_harmonised) |> arrange(n) |> print(n=99)
 
 

dev/composition.R

@@ -0,0 +1,154 @@
+library(tidyverse)
+library(forcats)
+library(HCAquery)
+library(dittoSeq)
+library(sccomp)
+
+source("https://gist.githubusercontent.com/stemangiola/fc67b08101df7d550683a5100106561c/raw/a0853a1a4e8a46baf33bad6268b09001d49faf51/ggplot_theme_multipanel")
+
+cell_metadata_with_harmonised_annotation = readRDS("dev/cell_metadata_with_harmonised_annotation.rds")
+
+data_for_plot_1 = 
+  cell_metadata_with_harmonised_annotation |> 
+  
+  left_join(
+    get_metadata("dev/metadata.SQLite") |> 
+      select(.cell, is_primary_data.y, name, cell_type, file_id, assay) |> 
+      as_tibble()
+  ) 
+
+# - Number of datasets per tissue
+plot_count_dataset = 
+  data_for_plot_1 |> 
+  distinct(file_id, tissue_harmonised) |> 
+  count(tissue_harmonised, name = "Number of datasets") |> 
+  ggplot(aes(fct_reorder(tissue_harmonised, desc(`Number of datasets`)), `Number of datasets`)) +
+  geom_bar(stat = "identity") +
+  xlab("Tissue") +
+  ylab("Number of datasets (log10)") +
+  scale_y_log10() +
+  theme_multipanel +
+  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5))
+
+# - Number of samples per tissue
+plot_sample_dataset = 
+  data_for_plot_1 |> 
+  distinct(.sample, tissue_harmonised) |> 
+  count(tissue_harmonised, name = "Number of samples") |> 
+  ggplot(aes(fct_reorder(tissue_harmonised, desc(`Number of samples`)), `Number of samples`)) +
+  geom_bar(stat = "identity") +
+  xlab("Tissue") +
+  ylab("Number of samples (log10)") +
+  scale_y_log10() +
+  theme_multipanel +
+  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5))
+
+# - Histogram of cells per sample
+plot_cell_dataset = 
+  data_for_plot_1 |> 
+  count(.sample, assay) |> 
+  ggplot(aes(n)) +
+  geom_histogram(aes(fill=assay), bins = 100) +
+  scale_fill_manual(values = dittoSeq::dittoColors()) +
+  xlab("Number of cells in sample (log10)") +
+  ylab("Count instances") +
+  scale_x_log10() +
+  theme_multipanel 
+
+# - Immune proportion per tissue
+data_for_immune_proportion = 
+  cell_metadata_with_harmonised_annotation |> 
+  
+  left_join(
+    get_metadata("dev/metadata.SQLite") |> 
+      select(.cell, is_primary_data.y, name, cell_type, file_id) |> 
+      as_tibble()
+  ) |> 
+  
+  # # Filter only whole tissue
+  # filter(
+  #   !name |> str_detect(regex('immune', ignore_case = T)) | 
+  #     tissue_harmonised %in% c("blood", "lymph node", "bone") |
+  #     is_primary_data.y == "PRIMARY"
+  # ) |> 
+  
+  # Filter Immune enriched dataset
+  filter(file_id != "e756c34a-abe7-4822-9a35-55ef12270247") |> 
+  filter(file_id != "ca4a7d56-739b-4e3c-8ecd-28704914cc14") |> 
+  filter(file_id != "59dfc135-19c1-4380-a9e8-958908273756" | tissue_harmonised != "intestine") |> 
+  
+  # nest(data = -c(.sample, tissue_harmonised)) |> 
+  # filter(map_int(data, ~ .x |> filter(cell_type_harmonised == "non_immune") |> nrow()) > 0 | tissue_harmonised %in% c("blood", "lymph node", "bone")) |> 
+  # unnest(data) |> 
+  
+  mutate(is_immune = cell_type_harmonised!="non_immune") |> 
+  
+  # Fix hematopoietic misclassificsation
+  mutate(is_immune = if_else(!is_immune & cell_type |> str_detect("hematopoietic"), TRUE, is_immune)) |> 
+  
+  # Filter out
+  filter(!cell_type |> str_detect("erythrocyte")) |> 
+  filter(!cell_type |> str_detect("platelet")) 
+
+data_for_immune_proportion_count = 
+  data_for_immune_proportion |> 
+  
+  # Stats
+  count(.sample, tissue_harmonised, is_immune, file_id) |> 
+  with_groups(.sample, ~ .x |> mutate(proportion = n/sum(n), sum = sum(n))) |> 
+  filter(is_immune) |> 
+  with_groups(tissue_harmonised, ~ .x |> mutate( median_proportion = mean(proportion)))
+
+dropLeadingZero <- function(l){  stringr::str_replace(l, '0(?=.)', '') }
+S_sqrt <- function(x){sign(x)*sqrt(abs(x))}
+IS_sqrt <- function(x){x^2*sign(x)}
+S_sqrt_trans <- function() scales::trans_new("S_sqrt",S_sqrt,IS_sqrt)
+
+
+plot_immune_proportion_dataset = 
+  data_for_immune_proportion_count |> 
+  ggplot(aes(fct_reorder(tissue_harmonised, desc(median_proportion)), proportion)) +
+  geom_point(aes(size = sum, color=file_id)) +
+  guides(color="none") +
+  scale_size(trans = "log10", range = c(0.1, 2.5), limits = c(1000, 10000)) +
+  scale_color_manual(values = dittoSeq::dittoColors()) +
+  scale_y_continuous(trans=S_sqrt_trans(), labels = dropLeadingZero) +
+  theme_multipanel +
+  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5))
+
+# - scatter plot of abundance vs variability per tissue
+
+
+# - Confidence class per cell type
+# - 
+
+# Study annotation
+res =
+  data_for_immune_proportion |>
+  mutate(is_immune = as.character(is_immune)) |> 
+  filter(tissue_harmonised %in% c("blood", "intestine")) |> 
+  sccomp_glm(
+    formula_composition = ~ 0 + tissue_harmonised,
+    formula_variability = ~ 1,
+    .sample, is_immune, 
+    check_outliers = FALSE,
+    approximate_posterior_inference = FALSE,
+    #contrasts = c("typecancer - typehealthy", "typehealthy - typecancer"),
+    cores = 20,
+    mcmc_seed = 42, verbose = T
+  )
+
+
+res |> plot_summary()
+
+cell_metadata_with_harmonised_annotation |> 
+  
+
+  
+  mutate(is_immune = cell_type_harmonised!="non_immune") |> 
+  
+  # Stats
+  count(.sample, tissue_harmonised, is_immune) |> 
+  with_groups(.sample, ~ .x |> mutate(proportion = n/sum(n))) |> 
+  filter(tissue_harmonised=="heart" & proportion > 0.75)
+  

dev/tissue_label_curated.csv

@@ -139,7 +139,7 @@ stomach,stomach
 testis,testis
 thymus,thymus
 thyroid gland,thyroid gland
-anterior part of tongue,tngue
+anterior part of tongue,tongue
 posterior part of tongue,tongue
 tongue,tongue
 trachea,trachea