Sample distribution plots: account for multiple samples from same individual

analyses/sample-distribution-analysis/01-filter-across-types.R

@@ -2,29 +2,29 @@
 # of key variables within the dataset.
 #
 # Chante Bethell for CCDL 2019
-# 
+#
 # #### USAGE
-# This script is intended to be run via the command line from the top directory 
+# This script is intended to be run via the command line from the top directory
 # of the repository as follows:
 #
 # Rscript analyses/sample-distribution-analysis/01-filter-across-types.R
 
 # magrittr pipe
 `%>%` <- dplyr::`%>%`
 
-# Function to filter based on primary_site 
+# Function to filter based on primary_site
 location_fn <- function(location) {
-  # Given the name of a primary site, create a vector containing the disease 
+  # Given the name of a primary site, create a vector containing the disease
   # types expressed within the primary site.
-  # 
-  # Note: the disease types found at all instances of the location substring 
-  #       will be included 
+  #
+  # Note: the disease types found at all instances of the location substring
+  #       will be included
   #
   # Args:
   #   location: the name of a primary site found within the dataset
   #
   # Returns:
-  #   disease_type_vector: the vector of disease types expressed at the 
+  #   disease_type_vector: the vector of disease types expressed at the
   #                        named primary site
   disease_type_vector <- brain_location %>%
     dplyr::arrange(dplyr::desc(n)) %>%
@@ -51,26 +51,34 @@ if (!dir.exists(plots_dir)) {
   dir.create(plots_dir)
 }
 
-# Read in dataset
-df2 <- data.frame(readr::read_tsv(
-  file.path(root_dir, "data", "pbta-histologies.tsv")
-))
-
-# Remove na's
-df2 <- df2 %>%
+# Read in dataset and remove NAs
+histologies_df <-
+  readr::read_tsv(file.path(root_dir, "data", "pbta-histologies.tsv")) %>%
+  as.data.frame() %>%
   dplyr::filter(!is.na(disease_type_new))
 
+# Filter the histologies file to account for multiple samples from the same
+# individual and the fact that multiple experimental strategies are in this
+# data.frame
+
+# Retain only tumors for this analysis
+histologies_df <- histologies_df %>%
+  dplyr::filter(sample_type == "Tumor",
+                composition == "Solid Tissue")
+
 # data.frame with the count of each unique cancer type expression
-disease_expression <- df2 %>%
+disease_expression <- histologies_df %>%
+  # some recurrences can have different disease_type_new values
+  dplyr::distinct(Kids_First_Participant_ID, disease_type_new) %>%
   dplyr::group_by(disease_type_new) %>%
   dplyr::count(name = "count") %>%
   dplyr::arrange(dplyr::desc(count))
 
 # Calculate the total count of the dataset
 sum_count <- sum(disease_expression$count)
 
-# Create a percent variable and round to 4 decimal places 
-# (so values will have 2 decimal places as percentages) 
+# Create a percent variable and round to 4 decimal places
+# (so values will have 2 decimal places as percentages)
 disease_expression <- disease_expression %>%
   dplyr::mutate(percent = paste0((round(count / sum_count, 4) * 100), "%"))
 
@@ -108,15 +116,17 @@ ggplot2::ggsave(
   height = 10
 )
 
-# data.frame with the location where each cancer type in the dataset is 
+# data.frame with the location where each cancer type in the dataset is
 # expressed, sorted to show highest expression
-brain_location <- df2 %>%
+brain_location <- histologies_df %>%
+  dplyr::distinct(Kids_First_Participant_ID, disease_type_new,
+                  primary_site) %>%
   dplyr::select(disease_type_new, primary_site) %>%
   dplyr::group_by(disease_type_new, primary_site) %>%
   dplyr::tally() %>%
   dplyr::arrange(dplyr::desc(n))
 
-# Make a vector of primary sites 
+# Make a vector of primary sites
 primary_sites_vector <- c(
   "Basal Ganglia",
   "Brain Stem- Midbrain",
@@ -143,11 +153,11 @@ primary_sites_vector <- c(
 # This step helps us with melting
 names(primary_sites_vector) <- primary_sites_vector
 
-# For each string in primary sites vector use location_fn to get the vector of 
-# disease types it's sorted by 
+# For each string in primary sites vector use location_fn to get the vector of
+# disease types it's sorted by
 cancer_types_list <- lapply(primary_sites_vector, location_fn)
 
-# Count the disease types for each primary site by taking the length of each 
+# Count the disease types for each primary site by taking the length of each
 # element of the list
 cancer_types_counts <- lapply(cancer_types_list, length)
 

analyses/sample-distribution-analysis/02-multilayer-plots.R

@@ -7,7 +7,7 @@
 # Chante Bethell for CCDL 2019
 #
 # #### USAGE
-# This script is intended to be run via the command line from the top directory 
+# This script is intended to be run via the command line from the top directory
 # of the repository as follows:
 #
 # Rscript analyses/sample-distribution-analysis/02-multilayer-plots.R
@@ -25,15 +25,19 @@ output_dir <- file.path(root_dir, "analyses", "sample-distribution-analysis")
 results_dir <- file.path(output_dir, "results")
 plots_dir <- file.path(output_dir, "plots")
 
-# Read in dataset 
-df2 <- readr::read_tsv(file.path(root_dir, "data",
+# Read in dataset
+histologies_df <- readr::read_tsv(file.path(root_dir, "data",
                                  "pbta-histologies.tsv"))
 
 # Create a colorblind-friendly color vector
 color <- colorblindr::palette_OkabeIto
 
 # Create final data.frame prepped for treemap and sunburst functions
-final_df <- df2 %>%
+final_df <- histologies_df %>%
+  dplyr::filter(sample_type == "Tumor",
+                composition == "Solid Tissue") %>%
+  dplyr::distinct(Kids_First_Participant_ID, broad_histology,
+                  short_histology, disease_type_new) %>%
   # Select our 3 columns of interest
   dplyr::select(broad_histology, short_histology, disease_type_new) %>%
   # Remove any row that has an NA
@@ -45,7 +49,7 @@ final_df <- df2 %>%
   # Place the value 1 in a column named counter for treemap and sunburt plots
   dplyr::mutate(counter= c(1)) %>%
   # Change the column names
-  dplyr::rename(level1 = broad_histology, 
+  dplyr::rename(level1 = broad_histology,
                 level2 = short_histology,
                 level3 = disease_type_new) %>%
   # Reorder the rows according to the 3 levels
@@ -56,7 +60,7 @@ final_df <- df2 %>%
 # Save to tsv file
 readr::write_tsv(final_df, file.path(results_dir, "plots_df.tsv"))
 
-# Create a treemap 
+# Create a treemap
 tm <-
   treemap::treemap(
     final_df,
@@ -67,19 +71,19 @@ tm <-
   )$tm
 
 # Convert the tm data.frame into a d3.js hierarchy object which is needed
-# for the sund2b plot 
+# for the sund2b plot
 tmnest <-
   d3r::d3_nest(tm[, c("level1", "level2", "level3", "vSize")],
                value_cols = c("vSize"))
 
-# Create an interactive treemap 
+# Create an interactive treemap
 interactive_tm <-
   d3treeR::d3tree(tm,
                   rootname = "Cancer Histologies Treemap",
                   width = 1200,
                   height = 700)
 
-# Create a sunburst plot 
+# Create a sunburst plot
 sun_plot <-
   sunburstR::sunburst(
     data = tmnest,
@@ -89,10 +93,10 @@ sun_plot <-
     colors = color
   )
 
-# Create an interactive sund2b plot 
+# Create an interactive sund2b plot
 p <- sunburstR::sund2b(tmnest, colors = color, valueField = "vSize")
 
-# Create HTML outputs for the interactive plots 
-mapview::mapshot(interactive_tm, url = file.path(plots_dir, 
+# Create HTML outputs for the interactive plots
+mapview::mapshot(interactive_tm, url = file.path(plots_dir,
                                                  "histology-treemap.html"))
 mapview::mapshot(p, url = file.path(plots_dir, "histology-pie.html"))

analyses/sample-distribution-analysis/results/disease_expression.tsv

@@ -1,74 +1,74 @@
 disease_type_new	count	percent
-Low-grade glioma;astrocytoma (WHO grade I/II)	498	24.88%
-High-grade glioma;astrocytoma (WHO grade III/IV)	230	11.49%
-Medulloblastoma	228	11.39%
-Ependymoma	173	8.64%
-Brainstem glioma- Diffuse intrinsic pontine glioma	130	6.49%
-Ganglioglioma	95	4.75%
-Craniopharyngioma	74	3.7%
-Atypical Teratoid Rhabdoid Tumor	60	3%
-Meningioma	59	2.95%
-Dysembryoplastic neuroepithelial tumor	50	2.5%
-Neurofibroma;Plexiform	39	1.95%
-Schwannoma	37	1.85%
-Choroid plexus papilloma	30	1.5%
-Supratentorial or Spinal Cord PNET	29	1.45%
-Dysplasia;Gliosis	27	1.35%
-Ewings Sarcoma	15	0.75%
-Teratoma	15	0.75%
-Chordoma	12	0.6%
-Metastatic secondary tumors	11	0.55%
-Pineoblastoma	10	0.5%
-Germinoma	8	0.4%
-Langerhans Cell histiocytosis	8	0.4%
-Adenoma	7	0.35%
-Choroid plexus carcinoma	7	0.35%
-Glial-neuronal tumor NOS	7	0.35%
-Malignant peripheral nerve sheath tumor	7	0.35%
-Sarcoma	7	0.35%
-Ganglioneuroblastoma	6	0.3%
-Hemangioblastoma	6	0.3%
-Meningioangiomatosis	6	0.3%
-Neuroblastoma	6	0.3%
-Neurocytoma	6	0.3%
-Subependymal Giant Cell Astrocytoma (SEGA)	6	0.3%
-Cyst	5	0.25%
-Dermoid Cyst	5	0.25%
-Gliomatosis Cerebri	4	0.2%
-Non-Langerhans Histiocytosis;JXG	4	0.2%
-Oligodendroglioma	4	0.2%
-Osteoblastoma	4	0.2%
-Rhabdomyosarcoma	4	0.2%
-Cavernoma	3	0.15%
-Brain arteriovenous malformation	2	0.1%
-Chondrosarcoma	2	0.1%
-Choroid plexus cyst	2	0.1%
-CNS embryonal tumor	2	0.1%
-Cortical Tubers	2	0.1%
-Dysembryoplastic neuroepithelial tumor (DNET);Ganglioglioma	2	0.1%
-Dysplasia;Gliosis;Glial-neuronal tumor NOS	2	0.1%
-Embryonal tumor with multilayer rosettes NOS	2	0.1%
-Ependymoblastoma	2	0.1%
-Fibroma	2	0.1%
-Fibromyxoid Tumor	2	0.1%
-Ganglioglioma;Low-grade glioma/astrocytoma (WHO grade I/II)	2	0.1%
-Ganglioneuroma	2	0.1%
-Germinoma;Teratoma	2	0.1%
-Hamartoma	2	0.1%
-Intraneural perineurioma	2	0.1%
-Malignant melanocytic neoplasm	2	0.1%
-Metastatic secondary tumors;Neuroblastoma	2	0.1%
-Myeloid Sarcoma	2	0.1%
-Myofibroblastoma	2	0.1%
-Myxoid spindle cell tumor	2	0.1%
-NeuroInflammatory systemic disease	2	0.1%
-Non-germinomatous germ cell tumor;Teratoma	2	0.1%
-Ossifying Fibroma	2	0.1%
-Primary CNS lymphoma	2	0.1%
-Reactive Connective Tissue	2	0.1%
-Reactive Gliosis	2	0.1%
-Rosai-Dorfman Disease	2	0.1%
-Dysembryoplastic neuroepithelial tumor (DNET);Dysplasia/Gliosis;Ganglioglioma	1	0.05%
-Medulloepithelioma	1	0.05%
-Papillary glioneuronal tumor	1	0.05%
-Prolactinoma	1	0.05%
+Low-grade glioma;astrocytoma (WHO grade I/II)	239	24.95%
+Medulloblastoma	119	12.42%
+High-grade glioma;astrocytoma (WHO grade III/IV)	101	10.54%
+Ependymoma	85	8.87%
+Ganglioglioma	46	4.8%
+Brainstem glioma- Diffuse intrinsic pontine glioma	45	4.7%
+Craniopharyngioma	39	4.07%
+Atypical Teratoid Rhabdoid Tumor	28	2.92%
+Meningioma	27	2.82%
+Dysembryoplastic neuroepithelial tumor	25	2.61%
+Neurofibroma;Plexiform	19	1.98%
+Choroid plexus papilloma	16	1.67%
+Schwannoma	16	1.67%
+Supratentorial or Spinal Cord PNET	16	1.67%
+Dysplasia;Gliosis	14	1.46%
+Teratoma	8	0.84%
+Ewings Sarcoma	7	0.73%
+Metastatic secondary tumors	5	0.52%
+Adenoma	4	0.42%
+Choroid plexus carcinoma	4	0.42%
+Cyst	4	0.42%
+Germinoma	4	0.42%
+Glial-neuronal tumor NOS	4	0.42%
+Langerhans Cell histiocytosis	4	0.42%
+Malignant peripheral nerve sheath tumor	4	0.42%
+Neuroblastoma	4	0.42%
+Pineoblastoma	4	0.42%
+Sarcoma	4	0.42%
+Chordoma	3	0.31%
+Dermoid Cyst	3	0.31%
+Meningioangiomatosis	3	0.31%
+Neurocytoma	3	0.31%
+Subependymal Giant Cell Astrocytoma (SEGA)	3	0.31%
+Cavernoma	2	0.21%
+Ganglioneuroblastoma	2	0.21%
+Gliomatosis Cerebri	2	0.21%
+Hemangioblastoma	2	0.21%
+Non-Langerhans Histiocytosis;JXG	2	0.21%
+Oligodendroglioma	2	0.21%
+Osteoblastoma	2	0.21%
+Rhabdomyosarcoma	2	0.21%
+Brain arteriovenous malformation	1	0.1%
+Chondrosarcoma	1	0.1%
+Choroid plexus cyst	1	0.1%
+CNS embryonal tumor	1	0.1%
+Cortical Tubers	1	0.1%
+Dysembryoplastic neuroepithelial tumor (DNET);Dysplasia/Gliosis;Ganglioglioma	1	0.1%
+Dysembryoplastic neuroepithelial tumor (DNET);Ganglioglioma	1	0.1%
+Dysplasia;Gliosis;Glial-neuronal tumor NOS	1	0.1%
+Embryonal tumor with multilayer rosettes NOS	1	0.1%
+Ependymoblastoma	1	0.1%
+Fibroma	1	0.1%
+Fibromyxoid Tumor	1	0.1%
+Ganglioglioma;Low-grade glioma/astrocytoma (WHO grade I/II)	1	0.1%
+Ganglioneuroma	1	0.1%
+Germinoma;Teratoma	1	0.1%
+Hamartoma	1	0.1%
+Intraneural perineurioma	1	0.1%
+Malignant melanocytic neoplasm	1	0.1%
+Medulloepithelioma	1	0.1%
+Metastatic secondary tumors;Neuroblastoma	1	0.1%
+Myeloid Sarcoma	1	0.1%
+Myofibroblastoma	1	0.1%
+Myxoid spindle cell tumor	1	0.1%
+NeuroInflammatory systemic disease	1	0.1%
+Non-germinomatous germ cell tumor;Teratoma	1	0.1%
+Ossifying Fibroma	1	0.1%
+Papillary glioneuronal tumor	1	0.1%
+Primary CNS lymphoma	1	0.1%
+Prolactinoma	1	0.1%
+Reactive Connective Tissue	1	0.1%
+Reactive Gliosis	1	0.1%
+Rosai-Dorfman Disease	1	0.1%