NEWS

---

pairedGSEA 1.13.1

Bug fixes

• paired_ora(): the combined (paired) analysis now uses genes tested for both expression and splicing as its universe; previously genes without a splicing result were included. Missing paired results are now filled in the paired columns instead of overwriting the splicing relative risk and enrichment score. Affects paired ORA results since 1.1.0 (Bioconductor 3.18); expression-only and splicing-only ORA are unchanged.
• paired_diff(): test = "Wald" no longer fails; the reduced model is only used for the likelihood ratio test.
• The reduced model for the likelihood ratio test now removes the group term by name instead of the first term of the design.
• paired_diff(): surrogate variable estimation is now seeded (sva_seed), so results are reproducible between runs.

Changes

• paired_diff(): the number of automatically estimated surrogate variables is capped so that the model keeps enough residual degrees of freedom (sva_min_residual_df). An optional upper limit can be set with sva_max_n; when set, the number is also capped at the square root of the number of samples. An exact request (sva_n) stops with an informative error when it is unsafe instead of being reduced silently.
• paired_diff(): SVA diagnostics (requested, estimated and retained surrogate variables, residual degrees of freedom) are stored in metadata(result)$pairedGSEA_sva.

pairedGSEA 1.7.0

• Ensure compatibility with updated msigdbr and new R version
• Fixed rare bug in aggregate_pvalue

pairedGSEA 1.1.2

• Fix bug where DEXSeq fails if there are NAs in the metadata

pairedGSEA 1.1.1

• Improved how surrogate variables are transferred in the design formulas

pairedGSEA 1.1.0

• paired_ora now adjusts gene background for analysis to reduce bias
• paired_ora now also does an ora combining the genes from the two analyses
• Added plotting mode paired = TRUE to plot_ora for the new paired_ora analysis type
• Added baseMean to gene-level aggregation output

pairedGSEA 1.0.0

• Version bump for Bioconductor release
• Minor edits to documentation

pairedGSEA 0.99.6

• Improved test coverage and depth
• Coerced paired_ora output to DataFrame from data.table
• Added value field to data man pages

pairedGSEA 0.99.5

• Remove non-exported man pages

pairedGSEA 0.99.4

• Remove filter_gene_sets option, as it is inherent in fgsea::ora
• Added test depth and coverage for paired_diff and paired_ora
• Fixed wrong storing location for splicing intermediate results

pairedGSEA 0.99.3

• Rewrote code base to remove tidyverse dependencies
• Added \code{...} and \link{...} where appropriate in documentation
• Added input parameter checks
• Reduced redundant input parameters from aggregate_pvalue
• Modularized paired_ora and plot_ora
• Added filter_gene_sets parameter to help users reduce gene set bias
• Increased test coverage and depth
• Moved data scripts to inst/script

pairedGSEA 0.99.2

• Implement limma as alternative analysis method
• Increased test coverage significantly
• Removed usage of deprecated purrr::when

pairedGSEA 0.99.1

• Updated vignette with a brief paragraph on the motivation of pairedGSEA.
• Updated NAMESPACE to include all imported packages. Suggested packages are added in notes.

pairedGSEA 0.99.0

• Submitted to Bioconductor