version: "1.0.0" name: bioconductor-geotcgadata description: Gene Expression Omnibus(GEO) and The Cancer Genome Atlas (TCGA) provide us with a wealth of data, such as RNA-seq, DNA Methylation, SNP and Copy number variation data. It's easy to download data from TCGA using the gdc tool, but processing these data into a format suitable for bioinformatics analysis requires more work. This R package was developed to handle these data. when_to_use: Use when: GeoTcgaData; Differential expression analysis (GeoTcgaData); ATAC-seq chromatin accessibility (GeoTcgaData). Not for: Requires R ≥ 4.2.0 and Bioconductor ≥ 3.16 user-invocable: false

GeoTcgaData

Workflows

Tcga Id Conversion And Normalization

Convert TCGA ENSEMBL gene IDs to gene symbols and normalize raw counts to FPKM or TPM.

library(GeoTcgaData)
 
# 1. Read user inputs
data(profile)
data(gene_cov)
 
# 2. Convert TCGA ENSEMBL gene IDs to gene symbols
result_sym <- id_conversion_TCGA(profile)
 
# 3. Convert raw counts to FPKM
lung_squ_count2 <- matrix(c(1,2,3,4,5,6,7,8,9), ncol=3)
rownames(lung_squ_count2) <- c("DISC1","TCOF1","SPPL3")
colnames(lung_squ_count2) <- c("sample1","sample2","sample3")
result_fpkm <- countToFpkm(lung_squ_count2, keyType = "SYMBOL", gene_cov = gene_cov)
 
# 4. Convert raw counts to TPM
result_tpm <- countToTpm(lung_squ_count2, keyType = "SYMBOL", gene_cov = gene_cov)

Input: A raw count matrix and a gene coverage data frame (`gene_cov`); Output: Converted gene symbols, FPKM, or TPM matrices.

Standard Workflow

Perform differential expression analysis on RNA-seq count data using a Wilcoxon test or other methods.

library(GeoTcgaData)
 
# 1. Prepare the count matrix and group vector
df <- matrix(rnbinom(400, mu = 4, size = 10), 25, 16)
df <- as.data.frame(df)
rownames(df) <- paste0("gene", 1:25)
colnames(df) <- paste0("sample", 1:16)
group <- sample(c("group1", "group2"), 16, replace = TRUE)
 
# 2. Run differential_RNA to identify differentially expressed genes (Method 1: using data frame)
result1 <- differential_RNA(counts = df, group = group, filte = FALSE, method = "Wilcoxon")
 
# 3. Run differential_RNA to identify differentially expressed genes (Method 2: using SummarizedExperiment)
colData <- S4Vectors::DataFrame(row.names = colnames(df), group = group)
data <- SummarizedExperiment::SummarizedExperiment(assays=S4Vectors::SimpleList(counts=as.matrix(df)), colData = colData)
result2 <- differential_RNA(counts = data, groupCol = "group", filte = FALSE, method = "Wilcoxon")

Input: A count matrix (or `SummarizedExperiment`) and sample group annotations; Output: A data frame of differential expression results.

Cnv Differential Analysis

Identify genes with differential copy number variations between groups.

library(GeoTcgaData)
 
# 1. Prepare the copy number score matrix and sample groups
aa <- matrix(sample(c(0, 1, -1), 200, replace = TRUE), 25, 8)
rownames(aa) <- paste0("gene", 1:25)
colnames(aa) <- paste0("sample", 1:8)
sampleGroup <- sample(c("A", "B"), ncol(aa), replace = TRUE)
 
# 2. Run differential_CNV to extract differential CNV genes
diffCnv <- differential_CNV(aa, sampleGroup)

Input: A gene-level copy number score matrix and a group vector; Output: A data frame of genes with differential copy number variations.

Dna Methylation Differential Analysis

Identify differentially methylated genes from CpG methylation data.

library(GeoTcgaData)
 
# 1. Prepare CpG methylation matrix, sample groups, and CpG-to-gene mapping
cpgData <- matrix(runif(2000), nrow = 200, ncol = 10)
rownames(cpgData) <- paste0("cpg", seq_len(200))
colnames(cpgData) <- paste0("sample", seq_len(10))
sampleGroup <- c(rep("group1", 5), rep("group2", 5))
names(sampleGroup) <- colnames(cpgData)
cpg2gene <- data.frame(cpg = rownames(cpgData), gene = rep(paste0("gene", seq_len(20)), 10))
 
# 2. Run differential_methy to identify differentially methylated genes
result <- differential_methy(cpgData, sampleGroup, cpg2gene = cpg2gene, normMethod = NULL)

Input: A CpG methylation beta-value matrix, sample group vector, and a CpG-to-gene mapping data frame; Output: A data frame of differentially methylated genes.

Geo Chip Data Preprocessing

Preprocess GEO microarray data by averaging duplicate gene symbols and resolving multi-mapped probe IDs.

library(GeoTcgaData)
 
# 1. Average expression values of different IDs for the same gene using gene_ave
file_gene_ave <- data.frame(Gene = c("MARCH1","MARC1","MARCH1"), GSM1 = c(2.9, 4.7, 8.1), GSM2 = c(3.9, 5.7, 7.1))
result_ave <- gene_ave(file_gene_ave, 1)
 
# 2. Resolve multi-mapped probe IDs by either assigning the expression to each gene using repAssign or deleting them using repRemove
input_file <- data.frame(aa = c("MARCH1 /// MMA", "MARC1"), bb = c("2.9", "4.7"), cc = c("3.9", "5.7"))
repAssign_result <- repAssign(input_file, " /// ")
repRemove_result <- repRemove(input_file, " /// ")

Input: A data frame of probe/gene expression values with potential duplicate or multi-mapped IDs; Output: A cleaned data frame with unique gene symbols.

When to Use

• To perform differential expression analysis on RNA-seq count data using Wilcoxon rank-sum tests via differential_RNA().
• To convert TCGA ENSEMBL gene IDs to gene symbols using id_conversion_TCGA().
• To normalize raw RNA-seq counts to FPKM or TPM using countToFpkm() and countToTpm().
• To identify differentially methylated genes from CpG methylation matrices using differential_methy().
• To preprocess GEO microarray data by averaging duplicate gene symbols with gene_ave() and resolving multi-mapped probe IDs with repAssign() or repRemove().

When NOT to Use

• For complex multi-factor experimental designs in RNA-seq differential expression, use DESeq2 or edgeR because differential_RNA() is designed for simple two-group comparisons.
• For advanced microarray normalization and linear modeling, use limma because GeoTcgaData focuses on basic ID averaging and multi-mapping resolution.

Data Requirements

• Input Format: Standard R matrix or data.frame, or a SummarizedExperiment object.
• Structure: Gene expression matrices should have genes/probes as rows and samples as columns.
• Normalization State: Raw counts are required for countToFpkm() and countToTpm(). Methylation data should be beta-values or similar continuous matrices.

Key Parameters

• counts (NULL): Input count matrix or SummarizedExperiment object in differential_RNA().
• group (NULL): Vector of sample group annotations in differential_RNA().
• filte (TRUE): Logical indicating whether to filter low-expression genes in differential_RNA().
• method ("Wilcoxon"): Statistical method for differential analysis in differential_RNA().
• keyType ("SYMBOL"): Type of gene identifier used in countToFpkm() or countToTpm().
• gene_cov (NULL): Gene coverage data frame used for length normalization in FPKM/TPM conversion.
• normMethod (NULL): Normalization method for methylation data in differential_methy().
• cpg2gene (NULL): Data frame mapping CpG probe IDs to gene symbols in differential_methy().

Best Practices

• Perform quality control on SNP data using SNP_QC() before running differential SNP analysis with differential_SNP().
• Use model = "gene" in methylation analysis to avoid bias related to the number of CpGs per gene.
• Ensure the gene coverage dataset gene_cov is loaded before running countToFpkm() or countToTpm().

Common Pitfalls

• Using raw counts directly in downstream analyses without normalization: Fix by converting counts to FPKM or TPM using countToFpkm() or countToTpm().
• Multi-mapped probe IDs causing duplicate rows: Fix by using repAssign() or repRemove() to handle probe IDs mapped to multiple genes.
• CpG number bias in methylation differential analysis: Fix by setting the model to "gene" instead of "cpg" in differential_methy().

Alternatives

• DESeq2 for robust parametric differential expression analysis of RNA-seq data.
• edgeR for differential expression analysis using empirical Bayes estimation.
• limma for linear modeling of microarray and RNA-seq data.
• ChAMP for comprehensive Illumina methylation integration and analysis.

Citations

• Young et al. 2010, Genome Biol (for gene ontology selection bias).
• Oshlack and Wakefield 2009, Biol Direct (for transcript length bias).

References

• Homepage: bioconductor.org/packages/GeoTcgaData
• Vignette: https://bioconductor.org/packages/release/bioc/vignettes/GeoTcgaData/inst/doc/GeoTcgaData.html