10-make_synapse_tables.Rmd

---
title: "Make Synapse tables for Ex10 proteomics"
author: "Michael Nestor (michael.nestor@pnnl.gov)"
date: "5/27/2021"
output: html_document
---

```{r helpers, echo=FALSE}
library(dplyr)
library(tidyr)
library(tibble)

library(amlresistancenetworks)
syn <- synapseLogin()

make_global_table <- function(crosstab.syn, metadata.syn,
                              tabname, parentId) {
  # get metadata
  path_to_metadata <- syn$get(metadata.syn)$path
  metadata <- read.table(path_to_metadata, header=T, sep="\t",
                         colClasses="character")
  metadata <- metadata %>% dplyr::rename(`PNNL ID` = SampleID.abbrev)
  
  # get crosstab
  path_to_crosstab <- syn$get(crosstab.syn)$path
  crosstab <- read.table(path_to_crosstab,
                         sep='\t', header=T, check.names=F)
  crosstab <- as.data.frame(crosstab)
  crosstab$Gene <- rownames(crosstab)
  crosstab <- crosstab %>% select(Gene, everything())
  
  # merge
  crosstab_long <- crosstab %>% 
    pivot_longer(cols = c(everything(), -Gene),
                 names_to='Sample', values_to='LogRatio') %>%
    filter(!is.na(LogRatio)) %>%
    left_join(metadata, by = c("Sample" ="PNNL ID"))
}

make_phospho_table <- function(crosstab.syn, metadata.syn,
                               tabname, parentId) {
  # get metadata
  path_to_metadata <- syn$get(metadata.syn)$path
  metadata <- read.table(path_to_metadata, header=T, sep="\t",
                         colClasses="character")
  metadata <- metadata %>% dplyr::rename(`PNNL ID` = SampleID.abbrev)
  
  # get crosstab
  path_to_crosstab <- syn$get(crosstab.syn)$path
  crosstab <- read.table(path_to_crosstab,
                         sep='\t', header=T, check.names=F)
  crosstab <- as.data.frame(crosstab)
  crosstab$SiteID <- rownames(crosstab)
  crosstab$Gene <- sub("^(.*)-.*$", "\\1", crosstab$SiteID)
  crosstab <- crosstab %>% select(Gene, SiteID, everything())
  
  # merge
  crosstab_long <- crosstab %>% 
    pivot_longer(cols = c(everything(), -Gene, -SiteID),
                 names_to='Sample', values_to='LogRatio') %>%
    filter(!is.na(LogRatio)) %>%
    left_join(metadata, by = c("Sample" ="PNNL ID"))
}

```

```{r main_loop}

library(readxl)
synapseDict <- read_xlsx("data/Ex10synapseDictionary.xlsx")
syn <- synapseLogin()

for (i in 1:nrow(synapseDict)) {
  
  tabName <- synapseDict$tabName[i]
  crosstab.syn <- synapseDict$crosstab.syn[i]
  metadata.syn <- synapseDict$metadata.syn[i]
  synid <- synapseDict$table.syn[i]
  
  if (grepl("Proteomics", tabName)) {
    tab <- make_global_table(crosstab.syn, metadata.syn)
  } else if (grepl("Phosphoproteomics", tabName)) {
    tab <- make_phospho_table(crosstab.syn, metadata.syn)
  }
  # synTableStore(tab = tab,
  #               tabname = "amlDrugCombos Molm14 Proteomics",
  #               parentId = parentId)
  synTableUpdate(tab, synid)
}
```