GSVA

Dependencies & Environment

Package-intrinsic requirements from the Bioconductor landing page — reproduce in any R environment.

• Version: 2.6.2 · Bioconductor: 3.23 · R: ≥ 4.6
• Imports: BiocGenerics, MatrixGenerics, S4Vectors, S4Arrays, HDF5Array, SparseArray, DelayedArray, IRanges, Biobase, SummarizedExperiment, GSEABase, Matrix, DelayedMatrixStats, BiocParallel, SingleCellExperiment, BiocSingular, SpatialExperiment, sparseMatrixStats, cli, memuse
• Install: BiocManager::install("GSVA")

When to Use

• Sample-Wise Pathway Scoring: Calculating sample-specific pathway enrichment scores from a gene expression matrix using gsva.
• Dimensionality Reduction: Transforming a high-dimensional gene-by-sample matrix into a lower-dimensional pathway-by-sample matrix for downstream analysis.
• Method-Specific Enrichment: Applying specific single-sample enrichment methods like PLAGE, z-score, or ssGSEA using plageParam, zscoreParam, or ssgseaParam.

When NOT to Use

• For Simple Differential Expression: For simple differential expression of individual genes, use edgeR because it is designed for gene-level count data modeling rather than gene set enrichment.
• For Purely Managing Gene Sets: For purely importing and managing gene set collections without analysis, use GSEABase because it provides dedicated classes like GeneSetCollection and functions like getGmt.

Data Requirements

• Expression Matrix: A normalized gene expression matrix, an ExpressionSet, or a SummarizedExperiment object.
• Gene Sets: A list object of gene identifiers or a GeneSetCollection object.

Key Parameters

• kcdf ("auto"): Kernel consensus density estimation; use "Gaussian" for continuous data or "Poisson" for integer counts.
• kcdfNoneMinSampleSize (200): Minimum sample size at which the empirical cumulative distribution function is estimated directly without a kernel.
• tau (1): Exponent defining the weight of the tail in the random walk.
• maxDiff (TRUE): Logical flag specifying whether to calculate the enrichment score as the magnitude difference between the largest positive and negative random walk deviations.
• absRanking (FALSE): Logical flag implying whether a modified Kuiper statistic is used.
• sparse (FALSE): Logical flag used when the input expression data is stored in a sparse matrix.
• minSize (1): Minimum size of gene sets to be considered.
• maxSize (Inf): Maximum size of gene sets to be considered.

Best Practices

• Match Identifiers: Ensure gene identifiers between the expression data matrix and the gene sets belong to the same standard nomenclature to allow mapIdentifiers to work.
• Set the Correct Kernel: Use kcdf="Poisson" when providing RNA-seq integer count data directly to gsvaParam.
• Import GMTs Safely: Use readGMT to import gene sets from GMT files, as it handles duplicated gene set names better than base R functions.

Common Pitfalls

• Providing Raw Counts with Gaussian Kernel: Running GSVA on raw RNA-seq counts without changing the kernel. Fix: Set kcdf="Poisson" in gsvaParam when using integer counts.
• Identifier Mismatch: Gene identifiers do not match between the expression matrix and gene sets. Fix: Provide data using specialized containers (like ExpressionSet or SummarizedExperiment) so GSVA can automatically map identifiers.
• Duplicated Gene Set Names: Duplicated gene set names in GMT files causing errors with standard import functions. Fix: Use the readGMT function which deduplicates gene set names by default.

Alternatives

• GSEABase: For foundational gene set data structures and basic GMT file importing.
• qusage: Provides alternative functions like read.gmt for reading gene sets.
• edgeR: For gene-level differential expression analysis using log-CPM units.

Citations

• Hänzelmann S, Castelo R, Guinney J. "GSVA: gene set variation analysis for microarray and RNA-seq data." BMC Bioinformatics, 2013.
• Barbie et al. "Systematic RNA interference reveals that oncogenic KRAS-driven cancers require TBK1." Nature, 2009.

References

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

<!-- biomate-cta -->

---

Run this on BioMate

This skill is the knowledge layer — when, why, and how to use gsva. To run this analysis on your own data with managed compute, automated QC, and reproducible outputs, use BioMate — free to start.

▶ Open gsva on BioMate →