---
title: "Raw TCGA data using Bioconductor's ExperimentHub"
author: "Sonali Arora"
date: "`r doc_date()`"
vignette: >
  %\VignetteIndexEntry{Raw TCGA data using Bioconductor's ExperimentHub}
  %\VignetteEngine{knitr::rmarkdown}
output: 
  BiocStyle::html_document
---


# DE analysis with Raw TCGA data using Bioconductor's ExperimentHub and DESeq2

TCGA re-processed RNA-Seq data from 9264 Tumor Samples and 741 normal samples
across 24 cancer types and made it available via [GSE62944](http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE62944) from GEO.
This data is also available as an ExpressionSet from ExperimentHub and
can be used for Differential Expression Analysis.

In the below example, we show how one can download this dataset from 
ExperimentHub.

```{r get-tcga}
library(ExperimentHub)
eh = ExperimentHub()
query(eh , "GSE62944")
```
One can then extract the data for this using
```{r download-tcga}
tcga_data <- eh[["EH1"]]
```
The different cancer types can be accessed using -

```{r cancer-types}
 head(phenoData(tcga_data)$CancerType)
```
Above we show only the top 6 Cancer subtypes.

##  Case Study 

We are interested in identifying the IDH1 mutant and IDH1 wild type
samples from TCGA's Low Grade Glioma Samples and then conducting a 
differential expression analysis using DESeq2

```{r lgg}
# subset the expression Set to contain only samples from LGG.
lgg_data <- tcga_data[, which(phenoData(tcga_data)$CancerType=="LGG")]

# extract the IDHI mutant samples
mut_idx <- which(phenoData(lgg_data)$idh1_mutation_found=="YES")
mut_data <- exprs(lgg_data)[, mut_idx]

# extract the IDH1 WT samples
wt_idx <- which(phenoData(lgg_data)$idh1_mutation_found=="NO")
wt_data <- exprs(lgg_data)[, wt_idx]

# make a countTable.
countData <- cbind(mut_data, wt_data)

# for DE analysis with DESeq2 we need a sampleTable
samples= c(colnames(mut_data), colnames(wt_data))
group =c(rep("mut",length(mut_idx)), rep("wt", length(wt_idx)))
coldata <- cbind(samples, group)
colnames(coldata) <- c("sampleName", "Group")
coldata[,"Group"] <- factor(coldata[,"Group"], c("wt","mut"))

# Now we can run DE analysis
library(DESeq2)
ddsMat <- DESeqDataSetFromMatrix(countData = countData,
                                 colData = DataFrame(coldata),
                                 design = ~ Group)

dds <- ddsMat
dds <- dds[ rowSums(counts(dds)) > 1, ]
dds <- DESeq(dds)
res <- results(dds) 
summary(res)

```

For a detailed RNASeq analysis see [Mike Love's RNASeq workflow](http://www.bioconductor.org/help/workflows/rnaseqGene/)


# sessionInfo()

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