index.Rmd

---
title: "sgRNA + CCLE - heatmap"
author: "Ray"
date: "`r format(Sys.time(), '%d %B, %Y')`"
output: html_document
params:
    workdir: "./"
    prefix: "genomeCRIPSR"
    tissue.type: "PANCREAS"
    mut.genes: "TP53"
    genealt.RDS: "./CRISPR_annot_allgenes_status.RDS"
    sgRNA.RDS: "./GenomeCRISPR_full05112017.RDS"
    ccle_tpm.RDS: "./CCLE_RNASeq_tpm_matched_CRISPR.RDS"
runtime: shiny
---

```{r, warning=FALSE, message=FALSE, echo=FALSE}
#rm(list = ls())
cleanMem <- function(n=10) { for (i in 1:n) gc() }

### Combine gene alteration status in CCLE data + CRISPR sgRNA data from genomeCRISPR database
library(shiny)
library(tidyverse)
library(pheatmap)
library(RColorBrewer)
source("helper.R")
options('download.file.method'='curl')

#folder <- params$workdir
#knitr::opts_chunk$set(echo = TRUE)
#knitr::opts_knit$set(root.dir = folder)

prefix <- params$prefix
# tissue <- params$tissue
tissue.type <- toupper(c("PANCREAS", "AML", "BREAST"))
mut.genes <- params$gene
genealt.RDS <- params$genealt.RDS
sgRNA.RDS <- params$sgRNA.RDS
ccle_tpm.RDS <- params$ccle_tpm.RDS
```


```{r ccle, echo=FALSE, message=FALSE, warning=FALSE}
### Input files
# genealt.RDS (CCLE mutational data)
# sgRNA.RDS (genomeCRISPR logFC data)
# ccle_tpm.RDS (CCLE expresion data)

genealt <- readRDS(genealt.RDS)
genealt$name <- gsub(" ", "-", genealt$name)
genealt$type <- gsub(" .*", "", genealt$type) %>% 
  gsub("^B$", "BURKITT", .)
# sgRNA
sgRNA <- readRDS(sgRNA.RDS)
# ccle_tpm
ccle_tpm <- readRDS(ccle_tpm.RDS)
```

```{r echo=FALSE}
### Filtering 
##- Genes 
    #+ genes that have mutation info in CCLE
    #+ genes in the sgRNA for plotting 
#- Samples 
    #+ Type (tissue)
```


```{r snow, echo=FALSE, message=FALSE, fig.width=12, fig.height=12}
# observe({
#      # periodically collect
#      invalidateLater(1000,session)
#      cleanMem()
#    })
cleanMem()

# Copy the chunk below to make a group of checkboxes
checkboxGroupInput("tissue.type", label = h4("Please select:"), 
                   choices = genealt$type %>% unique %>% sort %>% c("ALL", .), 
                   inline = TRUE, width='600px', 
                   selected = "AML")

selectizeInput("plot.genes", "Plot Genes (you can either select from below or type in your genes of interest)", 
	sgRNA$symbol %>% c("TP53", "KRAS", "NRAS", "PTEN", "EGFR", "MYC", "SMAD4", "CDKN1A", "CDKN2A", "IDH1", "IDH2", .) %>% unique, 
	selected = c("PCNA", "BRAF", "TP53", "MDM2"), multiple = TRUE, width="700px",
	options = list(maxOptions = 5, placeholder = 'e.g., TP53, MDM2'))

selectizeInput("mut.genes", "Gene mutations info: (again, type in your genes of interest)", 
		colnames(genealt)[-c(1:9)] %>% gsub("-.*", "", .) %>% c("TP53", "KRAS", "NRAS", "PTEN", "EGFR", "MYC", "SMAD4", "CDKN1A", "CDKN2A", "IDH1", "IDH2", .) %>% unique, 
		selected = c("TP53"), multiple = TRUE, width="700px",
		options = list(maxOptions = 10, placeholder = 'e.g., TP53, MDM2'))

renderPlot({
  prefix <- paste(input$tissue.type, collapse = "_")
  tissue.type <- input$tissue.type 
  # tissue.type <- c("AML", "BREAST", "NSCLC", "BL", "MELANOMA", "COLON", "LIVER", "OVARY")
  # tissue.type <- "ALL"
  tissue.type <- input$tissue.type 
  plot.genes <- input$plot.genes
  
  #mut.genes <- c("TP53")
  mut.genes <- input$mut.genes

  # how to bin/combine coloring for mutations - simple, merge, all
  mut.color.type <- "merge"


### Run from below 
mut.genes.col <- expand.grid(mut.genes, c("-change", "-type")) %>% 
  mutate(Var1 = as.character(Var1), 
         Var2 = as.character(Var2), 
         name = paste0(Var1, Var2)) %>% 
  left_join(tibble(Var1 = mut.genes), ., by="Var1") %>% 
  pull(name)


# Sort by the number of mutations in the record by decending order
# NAs in column '-change' are the splice_sites
col.idx <- match(mut.genes.col, colnames(genealt))
mut.genes.annot <- genealt[, c(1:9, col.idx)] %>% 
  mutate(tissue = gsub("BREST", "BREAST", tissue)) %>% 
  mutate(type = ifelse(type == "B", "BURKITT", type)) %>% 
  replace(is.na(.), "Splice_Site") #%>%

# filter based on tissue type
min.samp <- 2
if (tissue.type != "ALL"){
  final.annot <- mut.genes.annot %>% 
    filter(type %in% tissue.type) 
  if(nrow(final.annot) >= min.samp){
    print(paste("### Running for type:", tissue.type, "with", nrow(final.annot), "samples")) 
  }else{
    stop(paste("### samples for this type", tissue.type, "is less than", min.samp, "samples"))
  }
}else{
  prefix <- "ALL"
  final.annot <- mut.genes.annot
}

final.annot <- final.annot %>% 
  mutate(Sample_name = paste(cell, pub, type, sep="_")) %>% 
  mutate(Sample_name = make.unique(Sample_name)) %>% 
  mutate_at(vars(contains("-type")), funs(merge_mutation(., type=mut.color.type)))

plot.sgRNA <- tibble(symbol = plot.genes) %>% 
  left_join(., sgRNA, by="symbol") %>% 
  dplyr::select(Gene = symbol, final.annot$name)



### pheatmap
x2 <- plot.sgRNA %>% 
  replace(is.na(.), 0) %>% 
  set_names(c("Gene", final.annot$Sample_name)) %>% 
  convert_tibble_to_dataframe()

# Defines the vector of colors for the legend (it has to be of the same lenght of breaksList)
x2.min <- floor(min(x2, na.rm = TRUE))
x2.max <- ceiling(max(x2, na.rm = TRUE))

breaks = c(seq(x2.min, -0.51, length.out = 101), seq(-0.5, 0.5, length.out = 101), seq(0.51, x2.max, length.out=101))
breaks2 <- breaks
# color <- colorRampPalette(c("cyan1", "grey5", "red"))(length(breaks2))
color <- colorRampPalette(c("lightskyblue1", "grey5", "red"))(length(breaks2))
# color <- colorRampPalette(c("cadetblue1", "grey5", "red"))(length(breaks2))
breaks2[length(breaks)] <- max(x2.max, max(breaks))
breaks2[1] <- min(x2.min, min(breaks))

fontsize_row <- pheatmap_fontsize_row(x2)
fontsize_col <- pheatmap_fontsize_col(x2)
  
### create_annot_for_color
annotation_col <- final.annot %>% 
  dplyr::select(Sample_name, type, mut.genes.col) %>% 
  dplyr::select(-contains("-change")) %>% 
  #select_if(~ length(unique(.)) > 1) %>% 
  set_names(gsub("-type", "", colnames(.))) %>%
  convert_tibble_to_dataframe()
  
# create custom colors for each column_side_colors
# type.dat <- matrix(NA, nrow=nrow(annotation_col), ncol=ncol(annotation_col))
# colnames(type.dat) <- mut.genes
mycolors <- list()
for(i in 1:ncol(annotation_col)){
  cur.typename <- colnames(annotation_col)[i]
  cur.muttype <- annotation_col[, i]
  
  if(cur.typename == "type"){
    num.cur.muttype <- cur.muttype %>% unique %>% length
    type_col <- colorRampPalette(brewer.pal(n = num.cur.muttype, name = "Spectral"))(num.cur.muttype)
    names(type_col) <- cur.muttype %>% unique
    mycolors[[cur.typename]] <- type_col
  }else{
    mycolors[[cur.typename]] <- mut_color_predefined()[unique(c("WT", sort(cur.muttype)))]
  }
}
  
# 
pdf_height <- pheatmap_pdf_height(x2) + 1
pdf_width <- pheatmap_pdf_width(x2)

  pheatmap(x2, color=color,
           clustering_method = "ward.D2",
           annotation_col = annotation_col,
           annotation_colors = mycolors,
           breaks = breaks2,
           cluster_cols = TRUE,
           cluster_rows = TRUE,
           scale = "none", 
           show_rownames = TRUE,
           fontsize_row = fontsize_row,
           fontsize_col = fontsize_col,
           border_color = NA,
           main = paste0(prefix, "\n")
  )
#})
}, height=650)
```