TCGAbiolinks has provided a few functions to download and prepare data from GDC for analysis. This section starts by explaning the different downloads methods and the SummarizedExperiment object, which is the default data structure used in TCGAbiolinks, followed by some examples.

Downloading and preparing data for analysis

Data download: Methods differences

There are two methods to download GDC data using TCGAbiolinks:

  • client: this method creates a MANIFEST file and download the data using GDC Data Transfer Tool this method is more reliable but it might be slower compared to the api method.
  • api: this methods used the GDC Application Programming Interface (API) to downlaod the data. This will create a MANIFEST file and the data downloaded will be compressed into a tar.gz file. If the size and the number of the files are too big this tar.gz will be too big whicih might have a high probability of download failure. To solve that we created the files.per.chunk argument which will split the files into small chunks, for example, if chunks.per.download is equal to 10 we will download only 10 files inside each tar.gz.
Data prepared: SummarizedExperiment object

A SummarizedExperiment object has three main matrices that can be accessed using the SummarizedExperiment package):

  • Sample matrix information is accessed via colData(data): stores sample information. TCGAbiolinks will add indexed clinical data and subtype information from marker TCGA papers.
  • Assay matrix information is accessed via assay(data): stores molecular data
  • Feature matrix information (gene information) is accessed via rowRanges(data): stores metadata about the features, including their genomic ranges
Summarized Experiment: annotation information

When using the funciton GDCprepare there is an argument called SummarizedExperiment which defines the output type a Summarized Experiment (default option) or a data frame. To create a summarized Experiment object we annotate the data with genomic positions with last patch release version of the genome available. For legacy data (data aligned to hg19) TCGAbiolinks is using GRCh37.p13 and for harmonized data (data aligned to hg38) now it is using GRCh38.p7 (May 2017).

Unfortunately, some of the updates changes/remove gene symbols, change coordinates, etc. Which might introduce some loss of data. For example, if the gene was removed we cannot map it anymore and that information will be lost in the SummarizedExperiment.

If you set SummarizedExperiment to FALSE, you will get the data unmofied just as they are in the files and ad your own annotation.

Also, there are no updated for DNA methylation data. But the last metadata available can be found here: http://zwdzwd.github.io/InfiniumAnnotation

Related discussions: issue 91, issue 50

Search and download data from legacy database using GDC api method

In this example we will download gene expression data from legacy database (data aligned against genome of reference hg19) using GDC api method and we will show object data and metadata.

query <- GDCquery(project = "TCGA-GBM",
                  data.category = "Gene expression",
                  data.type = "Gene expression quantification",
                  platform = "Illumina HiSeq", 
                  file.type  = "normalized_results",
                  experimental.strategy = "RNA-Seq",
                  barcode = c("TCGA-14-0736-02A-01R-2005-01", "TCGA-06-0211-02A-02R-2005-01"),
                  legacy = TRUE)
GDCdownload(query, method = "api", chunks.per.download = 10)
data <- GDCprepare(query)
# Gene expression aligned against hg19.
datatable(as.data.frame(colData(data)), 
              options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
              rownames = FALSE)
# Only first 100 to make render faster
datatable(assay(data)[1:100,], 
              options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
              rownames = TRUE)

rowRanges(data)

Search and download data for two samples from database

In this example we will download gene expression quantification from harmonized database (data aligned against genome of reference hg38). Also, it shows the object data and metadata.

# Gene expression aligned against hg38
query <- GDCquery(project = "TCGA-GBM",
                  data.category = "Transcriptome Profiling",
                  data.type = "Gene Expression Quantification", 
                  workflow.type = "HTSeq - FPKM-UQ",
                  barcode = c("TCGA-14-0736-02A-01R-2005-01", "TCGA-06-0211-02A-02R-2005-01"))
GDCdownload(query)
data <- GDCprepare(query)
datatable(as.data.frame(colData(data)), 
              options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
              rownames = FALSE)

datatable(assay(data)[1:100,], 
              options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
              rownames = TRUE)

rowRanges(data)