expranno

expranno is an R package for downstream RNA-seq workflows built around Ensembl-ID expression matrices and sample metadata. It standardizes four core steps:

https://dai540.github.io/expranno/

• annotate_expr() for human or mouse gene annotation
• merge_expr_meta() for stable expression-metadata integration
• run_cell_deconvolution() for immunedeconv-based deconvolution
• run_signature_analysis() for GSVA or ssGSEA signature scoring

The package is intentionally narrow. It does not handle alignment, quantification, QC, or differential expression. Instead, it standardizes annotation presets, validation tables, provenance files, and downstream outputs once an expression matrix already exists.

It is designed for analysts working with bulk RNA-seq matrices where:

• expr: first column is gene_id, remaining columns are samples
• meta: first column is sample

For repeatable annotation, expranno ships fixed presets such as human_tpm_v102 and mouse_tpm_v102, plus bundled truth tables through example_annotation_truth().

expranno workflow

Installation

Install from GitHub:

install.packages("pak")
pak::pak("dai540/expranno")

Or:

install.packages("remotes")
remotes::install_github("dai540/expranno")

Or install from a source tarball:

install.packages("path/to/expranno_<version>.tar.gz", repos = NULL, type = "source")

Optional annotation and signature backends:

BiocManager::install(c(
  "biomaRt",
  "AnnotationDbi",
  "org.Hs.eg.db",
  "org.Mm.eg.db",
  "ensembldb",
  "EnsDb.Hsapiens.v86",
  "EnsDb.Mmusculus.v79",
  "GSVA"
))

Optional deconvolution backend:

remotes::install_github("omnideconv/immunedeconv")

Then load the package:

library(expranno)

Citation

If you use expranno, cite the package as:

Dai (2026). expranno: Expression Annotation, Metadata Integration, Deconvolution, and Signature Analysis. R package. https://dai540.github.io/expranno/

You can also retrieve the citation from R:

citation("expranno")

What expranno does

expranno does four things.

• Annotates Ensembl IDs with fixed-release human or mouse presets
• Benchmarks and validates annotation against truth tables
• Merges annotated expression with metadata into stable CSV outputs
• Runs deconvolution and signature scoring from the same annotated matrix

In practice, the package is doing this:

• annotate_expr() writes expr_anno.csv plus coverage, ambiguity, and provenance tables
• validate_annotation_engines() compares predicted fields against truth tables
• merge_expr_meta() writes expr_meta_merged.csv
• run_cell_deconvolution() writes cell_deconv_<method>.csv
• run_signature_analysis() writes signature_gsva.csv and signature_ssgsea.csv
• run_expranno() orchestrates the full workflow and keeps the output layout stable

Main functions

• run_expranno()
• annotate_expr()
• benchmark_annotation_engines()
• validate_annotation_engines()
• merge_expr_meta()
• run_cell_deconvolution()
• run_signature_analysis()
• as_expranno_input()
• as_expranno_se()
• list_annotation_presets()
• example_annotation_truth()

Stable outputs

• expr_anno.csv
• annotation_report.csv
• annotation_ambiguity.csv
• annotation_provenance.csv
• annotation_validation_summary.csv
• annotation_validation_detail.csv
• expr_meta_merged.csv
• cell_deconv_<method>.csv
• signature_gsva.csv
• signature_ssgsea.csv
• session_info.txt

Example

library(expranno)

demo <- example_expranno_data()

result <- run_expranno(
  expr = demo$expr,
  meta = demo$meta,
  annotation_preset = "human_tpm_v102",
  expr_scale = "abundance",
  duplicate_strategy = "mean",
  output_dir = tempdir(),
  run_deconvolution = FALSE,
  run_signature = FALSE
)

result$annotation$report

Tutorials

The tutorial site is organized around:

• Getting Started
• preset reference
• benchmarking and reproducibility
• human case study
• mouse case study
• function reference

What expranno cannot do yet

• It does not perform alignment, quantification, or read-level QC
• It does not replace native immunedeconv or GSVA method-specific tuning
• It does not provide a CRAN or Bioconductor release yet
• It does not implement a full reporting layer beyond CSV outputs, vignettes, and pkgdown docs

Package layout

• R/annotate.R: annotation workflows and backends
• R/run.R: end-to-end wrapper
• R/signature.R: GSVA and ssGSEA scoring
• R/deconvolution.R: deconvolution wrapper
• R/benchmark.R and R/validation.R: benchmark and truth-table validation
• R/bioconductor.R: Bioconductor input/output helpers
• R/data-helpers.R: built-in examples, truth tables, and preset listing
• R/io.R: strict input validation and expression-metadata merging
• vignettes/: getting-started, design, and case-study articles
• inst/extdata/: bundled demo GMT and truth tables