OncoSeq

End-to-end analysis of cancer transcriptomes and genomes in R.

OncoSeq keeps bulk RNA-seq counts, somatic mutations, gene-level copy number and clinical follow-up for one set of tumours in a single container, and provides the analyses a cancer genomics study needs from import to figure. Every result is stored back in the object, together with a provenance log, so a complete study travels as one file.

Installation

if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install("babasaraki/OncoSeq")

The TCGA importer additionally needs TCGAbiolinks:

BiocManager::install("TCGAbiolinks")

Quick start

library(OncoSeq)

# A curated cBioPortal study, or your own counts with importCounts()
ose <- importCBioPortal("blca_tcga_pan_can_atlas_2018",
                        genes = c("TP53", "RB1", "FGFR3", "ERBB2"))

# ... or the shipped example cohort
ose <- exampleOncoSeq()

ose <- qcMetrics(ose)
ose <- filterLowExpression(ose, group = "condition")
ose <- normalizeCounts(ose, method = "vst")
ose <- diffExpress(ose, ~ condition,
                   contrast = c("condition", "tumour", "normal"),
                   engine = "DESeq2")

plotVolcano(ose)
head(getAnalysis(ose, "de"))

Or run the whole chain at once:

res <- runOncoSeq(ose, ~ condition,
                  contrast = c("condition", "tumour", "normal"),
                  geneSets = readGMT("hallmark.gmt"))
names(analyses(res))

What is in the box

Stage	Functions
Import	importGDC(), importCBioPortal(), importCounts(), importTximport(), importMAF(), asOncoSeq()
Quality control	qcMetrics(), filterLowExpression(), detectOutliers()
Normalisation	normalizeCounts() — CPM, TMM, upper-quartile, RLE, VST, rlog, log2-TPM
Differential expression	diffExpress() over DESeq2 / edgeR / limma-voom, compareEngines()
Enrichment	runGSEA(), runORA(), readGMT(), msigdbrSets()
Mutations	mutationSummary(), mutationFrequency(), tumourMutationBurden(), oncoplotData()
Copy number	cnvSummary(), cnvFrequency(), cnvExpressionCorrelation()
Survival	survivalFit(), coxModel(), survivalScreen()
Integration	mutationStatus(), integrateMutationExpression(), alignLayers()
Plots	plotSamplePCA(), plotVolcano(), plotMAplot(), plotExpressionHeatmap(), plotKM(), plotOncoprint(), plotEnrichmentBar(), plotCNVFrequency(), plotMutationFrequency()
Pipeline	runOncoSeq()

The container

OncoSeqExperiment extends SummarizedExperiment, so assay(), colData(), rowData() and friends work as usual. It adds:

• mutations(x) — somatic variants in long format
• cnv(x) — gene-level copy-number calls
• analyses(x) — every result you have computed
• provenance(x) — what was done to the object, in order

Subsetting samples propagates to every layer, and validity checking keeps the sample identifiers concordant.

Data

• exampleOncoSeq() — a simulated count-scale cohort for examples and teaching
• blcaExample — a real 420-gene by 400-sample slice of the TCGA bladder urothelial carcinoma cohort, with expression, mutations, copy number and survival

Documentation

vignette("OncoSeq")

Licence

GPL-3.

Provenance

This package was independently conceived, designed, developed, and maintained by the author. AI tools, particularly Anthropic Claude Opus 5, were used selectively to provide technical assistance in areas where additional guidance was needed, including troubleshooting, debugging, code refinement, and resolving specific implementation challenges.

The author retained full control over the package's scope, architecture, functionality, source code, documentation, unit tests, and validation processes. All AI-assisted suggestions were critically reviewed, adapted where necessary, and tested by the author. Package validation included running R CMD check --as-cran and BiocCheck locally and through continuous integration.

No code was knowingly copied from third-party packages or repositories, and all dependencies are accessed through their publicly documented APIs. The package is distributed under the GPL-3 licence.

The author assumes full responsibility for the originality, correctness, maintenance, and scientific integrity of the package.

This disclosure is provided in accordance with Bioconductor's AI and Third Party Code policy.