---
title: "Converting PharmacoSet Drug Response Data into gDR object" 
author:
- name: Jermiah Joseph
  email: jermiah.joseph@uhn.ca
- name: Bartosz Czech
  email: bartosz.w.czech@gmail.com
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Converting PharmacoSet Drug Response Data into gDR object}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---


```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

```{r setup}
library(PharmacoGx)
library(gDRimport)
```


# Overview
The `gDRimport` package is a part of the gDR suite. It helps to prepare raw drug response data for downstream processing. 
It mainly contains helper functions for importing/loading/validating dose response data provided from different scanner sources. 
In collaboration with the BHKLab, `gDRimport` also provides functions that can convert a `PharmacoGx::PharamcoSet` object into a gDR object. 
With this functionality, users familiar with the gDR suite of packages and methods can utilize the publically available, curated datasets from the PharmacoGx database. 
The main step in this process is to extract the drug dose-response data from the PharmacoSets and transform them into a `data.table` that can be used as input for the `gDRcore::runDrugResponseProcessingPipeline`.


# Loading a PharmacoSet (PSet)
Whereas a user might already have a pharmacoset loaded in their R session, if they wish to obtain a different pharmacoset or use the same script in the future we provide a helper function to do so. 
It helps to have a user directory in which to store all pharmacosets, and by passing this directory into the function as a parameter, the function will also check to see if the PSet exists in the user-defined directory.
This is to ensure that the PSet is not being re-downloaded if it already has.  
```{r eval = FALSE}
pset <- getPSet("Tavor_2020")
pset
```

```{r, include = FALSE}
read_mocked_PSets <- function(canonical = FALSE) {
  qs::qread(
    system.file("extdata", "data_for_unittests", "PSets.qs", package = "gDRimport")
  )
}
pset <- testthat:::with_mock(
    `PharmacoGx::availablePSets` = read_mocked_PSets,
    suppressMessages(getPSet(
      "Tavor_2020", psetDir = system.file("extdata/pset", package = "gDRimport")
    ))
  )
pset
```

# Converting PharmacoSet to data.table for gDR pipeline
`PharamcoSets` hold data pertaining to the cell lines (@sample slot), drugs (@treatment slot), and dose response experiments (@treatmentResponse slot). 
The dose response data is stored in a `treatmentResponseExperiment` object and the function `gDRimport::convert_pset_to_df` extracts this information to 
build a `data.table` that can be used as input to the gDR pipeline. 

```{r}
# Store treatment response data in df_ 
dt <- convert_pset_to_df(pharmacoset = pset)
str(dt)
```

# Subsetting to extract relevant information
Most canonical `PharmacoSets` have data pertaining to many cell lines and their response to many drugs (drug-combination data is available in some but its conversion to gDR is not currently supported).
As such, in the interest of time and resources, it may be useful to subset the data before providing it as input for the gDR pipeline. 

```{r}
# example subset using only 1 cell line
subset_cl <- dt$Clid[1]
x <- dt[Clid == subset_cl]
x
```

# Running drug response pipeline with data
The subsetted data can now be used as input for the `gDRcore::runDrugResponseProcessingPipeline()`. The output of this function is a `MultiAssayExperiment` object which can be accessed with `gDRutils::convert_se_assay_to_dt()`

```{r eval = FALSE}
# RUN DRUG RESPONSE PROCESSING PIPELINE
se <- gDRcore::runDrugResponseProcessingPipeline(x)
se
``` 

```{r eval = FALSE}
# Convert Summarized Experiments to data.table 
# Available SEs : "RawTreatred", "Controls", "Normalized", "Averaged", "Metrics"

str(gDRutils::convert_se_assay_to_dt(se[[1]], "Averaged"))
str(gDRutils::convert_se_assay_to_dt(se[[1]], "Metrics"))
```


```{r, include = FALSE}
gDRutils::reset_env_identifiers()
```


# SessionInfo {-}

```{r sessionInfo}
sessionInfo()
```