supplemental_summary_table.Rmd

---
title: "Manuscript supplemental table"
author: "Camilo Posso"
date: "08/08/2022"
output: 
  html_document:
    code_folding: hide
    toc: true
editor_options: 
  chunk_output_type: inline
---

## Goal

Predict subtypes on the 51 held out samples, as well as entirely new datasets. CHeck
survival, mutation, etc.


```{r include=FALSE}
library(dplyr)
library(ggplot2)

source("../util/loading_data.R")
source("../util/synapseUtil.R")
source("../util/make_plots_util.R")
source("../util/mutational_analysis_helper.R")

syn <- synapseLogin()
meta <- load.metadata()
RNA.data <- load.RNA.data()
RNA_samples <- unique(RNA.data$Barcode.ID)

prediction_210 <- read.table(syn$get("syn30030154")$path, sep = "\t") %>%
  mutate(Cluster = paste("Predicted", Cluster))

all_mutation_data <- load_mutational_sample_data()
mutation_data <- all_mutation_data %>%
  mutate(value = TRUE) %>%
  pivot_wider(names_from = Gene, values_from = value, values_fn = any)
mutation_data <- mutation_data[, c("Barcode.ID", "KRAS+NRAS", "NPM1_clinical")] %>%
  full_join(meta[, c("Barcode.ID", "FLT3.ITD")], by = "Barcode.ID") %>%
  mutate(RAS = case_when(!is.na(`KRAS+NRAS`) ~ `KRAS+NRAS`,
                          TRUE ~ FALSE),
         NPM1_clinical = case_when(!is.na(NPM1_clinical) ~ NPM1_clinical,
                                   TRUE ~ FALSE),
         FLT3 = case_when(FLT3.ITD == "TRUE" ~ TRUE,
                          TRUE ~ FALSE)) %>%
  select(Barcode.ID, FLT3, NPM1_clinical, RAS)

rownames(mutation_data) <- mutation_data$Barcode.ID

# extra_data_cols <- c("SampleID.abbrev", "specimen.type", "Sepcimen.access.group.concatenated", "FLT3.ITD", "NPM1", "RAS", 
# "Subtype", "data_types", "specimen.location")
supplemental_table <- meta %>%
  select(Barcode.ID, Sample = SampleID.abbrev, 
         Location = specimen.location, 
         `Specimen Type` = specimen.type, 
         `Specimen Group` = Specimen.access.group.concatenated) %>%
  mutate(`Transcriptomics available` = Barcode.ID %in% RNA_samples) %>%
  merge(prediction_210, by = "Barcode.ID") %>%
  merge(mutation_data, by = "Barcode.ID") %>%
  dplyr::rename(Subtype = Cluster) %>%
  mutate(Subtype = sub("Predicted Cluster ", "Subtype ", Subtype),
         Sample = as.numeric(Sample)) %>%
  arrange(Sample) %>%
  select(-Barcode.ID) 

write.table(supplemental_table, "supplemental_table_1.txt", sep = "\t", quote = F, row.names = F)

```