TCGAbiolinks：搜索GDC数据库

发表时间：2019年3月20日

TCGAbiolinks提供了一些搜索GDC数据库的功能。本节首先介绍GDC数据的两个可以用的源：Harmonized和Legacy Archive。然后提供一些实例，探究如何如何访问和使用它们。

1. 一些有用的信息

1.1. 不同GDC源：Harmonized和Legacy Archive

TCGAbiolinks可以使用两个GDC源下载GDC数据：

• **GDC Legacy Archive：**提供对以前存储在CGHub和TCGA数据协调中心（DCC）并由TCGA Data Portal托管的未经修改的数据副本的访问，这里的数据使用GRCh37（hg19）和GRCh36（hg18）基因组作为参考。
• GDC harmonized database：这里的数据以GRCh38（hg38）基因组作为参考，并使用GDC Bioinformatics Pipelines对提供的生物样本和临床数据进行标准化

1.2. 理解Barcode

TCGA Barcode 由一组标识符组成。每一个Barcode都对应一个TCGA数据元素。有关元数据标识符如何构成Barcode 的说明，请参阅下面信息。其中，Aliquot Barcode包含最多数量的标识符。

案例：

• Aliquot barcode：TCGA-G4-6317-02A-11D-2064-05
• Participant：TCGA-G4-6317
• Sample：TCGA-G4-6317-02

有关更多信息，请查看TCGA wiki

2. 搜索参数

可以使用GDCquery函数轻松搜索GDC数据。该函数可以使用以下参数来搜索：

2.1. project options

在project中可以选择的选项如下：

表格内容注释：

• dbgap_accession_number：dbgap关联ID
• disease_type：疾病的类型
• releasable：是否发布
• released：是否公开
• state：目前的状态是否开放
• primary_site：样本组织来源
• project_id：项目的ID
• id：数据编号
• name：疾病名称
• tumor：疾病名称的缩写

2.2. sample.type options

sample.type可用参数如下：

其他搜索参数（data.category、data.type、workflow.type、platform、file.type）可以在下面找到。请注意，这些表格并不是完整的

2.3. Harmonized data options (legacy = FALSE)

datatable(readr::read_csv("https://docs.google.com/spreadsheets/d/1f98kFdj9mxVDc1dv4xTZdx8iWgUiDYO-qiFJINvmTZs/export?format=csv&gid=2046985454"),
          filter = 'top',
          options = list(scrollX = TRUE, keys = TRUE, pageLength = 40), 
          rownames = FALSE)
## Parsed with column specification:
## cols(
##   Data.category = col_character(),
##   Data.type = col_character(),
##   `Workflow Type` = col_character(),
##   Platform = col_character()
## )

2.4. Legacy archive data options (legacy = TRUE)

datatable(readr::read_csv("https://docs.google.com/spreadsheets/d/1f98kFdj9mxVDc1dv4xTZdx8iWgUiDYO-qiFJINvmTZs/export?format=csv&gid=1817673686"),
          filter = 'top',
          options = list(scrollX = TRUE, keys = TRUE, pageLength = 40), 
          rownames = FALSE)
## Parsed with column specification:
## cols(
##   Data.category = col_character(),
##   Data.type = col_character(),
##   Platform = col_character(),
##   file.type = col_character()
## )

3. Harmonized数据库案例

3.1. DNA甲基化数据：Recurrent tumor samples

在这个例子中，我们将访问Harmonized数据库（legacy = FALSE），并搜索recurrent glioblastoma multiform (GBM) 和low grade gliomas (LGG) 样本的所有DNA甲基化数据。

# 导入必备的包，在后续的代码中都要先导入包，才能使用这些函数
library(TCGAbiolinks)
library(SummarizedExperiment)

# 查询信息
query <- GDCquery(project = c("TCGA-GBM", "TCGA-LGG"),
                  data.category = "DNA Methylation",
                  legacy = FALSE,
                  platform = c("Illumina Human Methylation 450"),
                  sample.type = "Recurrent Solid Tumor"
)

# 展示查询结果
datatable(getResults(query), 
          filter = 'top',
          options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
          rownames = FALSE)

数据在这里只显示了一部分

3.2. DNA甲基化数据+基因表达数据：Colon Adenocarcinoma tumor

在这个例子中，我们将访问Harmonized数据库（legacy = FALSE），并搜索具有DNA甲基化数据和基因表达数据的Colon Adenocarcinoma tumor (TCGA-COAD)患者

query.met <- GDCquery(project = "TCGA-COAD",
                      data.category = "DNA Methylation",
                      legacy = FALSE,
                      platform = c("Illumina Human Methylation 450"))
query.exp <- GDCquery(project = "TCGA-COAD",
                      data.category = "Transcriptome Profiling",
                      data.type = "Gene Expression Quantification", 
                      workflow.type = "HTSeq - FPKM-UQ")

# Get all patients that have DNA methylation and gene expression.
common.patients <- intersect(substr(getResults(query.met, cols = "cases"), 1, 12),
                             substr(getResults(query.exp, cols = "cases"), 1, 12))

# Only seelct the first 5 patients
query.met <- GDCquery(project = "TCGA-COAD",
                      data.category = "DNA Methylation",
                      legacy = FALSE,
                      platform = c("Illumina Human Methylation 450"),
                      barcode = common.patients[1:5])
query.exp <- GDCquery(project = "TCGA-COAD",
                      data.category = "Transcriptome Profiling",
                      data.type = "Gene Expression Quantification", 
                      workflow.type = "HTSeq - FPKM-UQ",
                      barcode = common.patients[1:5])

datatable(getResults(query.met, cols = c("data_type","cases")),
          filter = 'top',
          options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
          rownames = FALSE)

datatable(getResults(query.exp, cols = c("data_type","cases")), 
          filter = 'top',
          options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
          rownames = FALSE)

3.3. 原始测序数据：

这个例子显示了用户如何搜索乳腺癌（Breast Cancer）的原始测序数据（“Controlled”），并验证文件名称和与之相关的Barcodes。

query <- GDCquery(project = c("TCGA-BRCA"),
                  data.category = "Sequencing Reads",  
                  sample.type = "Primary solid Tumor")
# Only first 5 to make render faster
datatable(getResults(query, rows = 1:5,cols = c("file_name","cases")), 
          filter = 'top',
          options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
          rownames = FALSE)

4. Legacy archive数据库案例

4.1. 甲基化数据

4.1.1. 甲基化数据 - 来自Array-based assays

该案例显示了用户如何搜索基于Illumina Human Methylation 450和Illumina Human Methylation 27平台的glioblastoma multiform (GBM)和low grade gliomas (LGG)的DNA甲基化数据。

query <- GDCquery(project = c("TCGA-GBM","TCGA-LGG"),
                  legacy = TRUE,
                  data.category = "DNA methylation",
                  platform = c("Illumina Human Methylation 450", "Illumina Human Methylation 27"))
datatable(getResults(query, rows = 1:100), 
          filter = 'top',
          options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
          rownames = FALSE)

数据在这里只显示了一部分

4.1.2. 甲基化数据 - 来自whole-genome bisulfite sequencing (WGBS)

query <- GDCquery(project = c("TCGA-LUAD"),
                  legacy = TRUE,
                  data.type = "Methylation percentage",
                  experimental.strategy = "Bisulfite-Seq")

# VCF - controlled data
query <- GDCquery(project = c("TCGA-LUAD"),
                  legacy = TRUE,
                  data.type = "Bisulfite sequence alignment",
                  experimental.strategy = "Bisulfite-Seq")


# WGBS BAM files - controlled data
query <- GDCquery(project = c("TCGA-LUAD"),
                  legacy = TRUE,
                  data.type = "Aligned reads",
                  data.category = "Raw sequencing data",
                  experimental.strategy = "Bisulfite-Seq")

4.2. 基因表达数据

该案例显示了用户如何搜索glioblastoma multiform (GBM)的基因表达数据，并且该数据已经进行基因表达量标准化。更详细的信息，请参考rnaseqV2 TCGA wiki

# Gene expression aligned against hg19.
query.exp.hg19 <- 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)
datatable(getResults(query.exp.hg19), 
          filter = 'top',
          options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
          rownames = FALSE)

5. 获得查询结果文件清单

如果要从查询对象获取文件清单, 可以使用函数getManifest。如果将函数中的save参数设置为 TRUE, 则将创建一个 txt 文件。同时该文件可以用于GDC-client Data transfer tool (DTT) 及其对应的GUI版本ddt-ui。

getManifest(query.exp.hg19,save = FALSE) 
##                                      id
## 40 217d72e9-4d6f-409d-911c-0a70b17a0adc
## 97 973ce0ac-f613-4b99-b2ab-3e2d5548f05f
##                                                                              filename
## 40 unc.edu.b469eb7c-723f-4870-b4e4-ebfaae7a118b.1536566.rsem.genes.normalized_results
## 97 unc.edu.152afe8c-f67c-4d7c-93ac-e1b7edd56c54.1544649.rsem.genes.normalized_results
##                                 md5   size state
## 40 beda9f89f08fc6a892a72e8b704fdbd9 437283  live
## 97 84478e78d95e1155019ccb7e0e0fea2f 436272  live

6. ATAC-seq数据

目前，ATAC-seq数据可以从GDC publication page获得，具体的列表如下：

datatable(getResults(TCGAbiolinks:::GDCquery_ATAC_seq())[,c("file_name","file_size")], 
          filter = 'top',
          options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
          rownames = FALSE)

我们还可以使用函数GDCquery_ATAC_seq过滤文件清单，并使用函数GDCdownload过滤后的数据保存到本地。

query <- TCGAbiolinks:::GDCquery_ATAC_seq(file.type = "rds") 
GDCdownload(query,method = "client")

query <- TCGAbiolinks:::GDCquery_ATAC_seq(file.type = "bigWigs") 
GDCdownload(query,method = "client")

7. 每位患者的信息

检索每个data_category + data_type + experimental_strategy + platform下的文件数。几乎像https://portal.gdc.cancer.gov/exploration

tab <-  getSampleFilesSummary("TCGA-ACC")
datatable(tab,
          filter = 'top',
          options = list(scrollX = TRUE, keys = TRUE, pageLength = 5), 
          rownames = FALSE)