Skip to contents

Changelog

Source: NEWS.md

geneSetSimplifyR 0.99.0

First version submitted to Bioconductor.

New features

  • Pluggable clustering methods: clusterGeneSets(algorithm =) accepts a registered name (‘louvain’, ‘louvain.multilevel’, ‘slm’, ‘leiden’, ‘igraph_louvain’, ‘igraph_leiden’, ‘kmeans’, ‘apcluster’, ‘binary_cut’), a method made with clusterMethod(), or a function. See registerClusterMethod(), listClusterMethods(), getClusterMethod(), snnToIgraph() and snnToMatrix().
  • clusterGeneSets(K =) sets the number of nearest neighbors; the default is round(sqrt(N)) for N gene sets.
  • Parsers for enrichment results: parseFgsea() (GSEA and ORA results) and parsePairedGSEA(). getGeneSets() fetches MSigDB gene sets.
  • initializeList() accepts a pval (or pvalue) column and adds nlog10pval. Other numeric columns are kept.
  • initializeList() and geneSetSimplifyR() accept the path to a GMT file as geneSetsList.
  • plotEnrichments(), plotHeatmap() and plotViolin() gain plotFrom (the column to plot; by default -log10 p-values when available, else enrichment scores). plotEnrichments() and plotHeatmap() gain summaryStat (default ‘median’).
  • plotHeatmap(scaleCols =) with default "auto": clusters are z-scored across sources only when there are three or more sources.
  • labelClusters(removeFirstWord =) drops the first word (e.g. an author name or database prefix) of gene-set names before labeling.
  • geneSetSimplifyR() passes K, stopWords and nLabels on.
  • The example data exampleEnrichment (and exampleGsList) include the fgsea p-values and adjusted p-values (pval, padj).
  • normalizeClusterLevels() updates objects clustered by earlier versions.
  • The gsList class has a metadata slot recording parameters and results.

Changes in behavior

  • With resolution = NULL, plotting and table functions use the active resolution: the optimal one from clusterGeneSets(), or the one last set with updateResolution() or updateClusterLabels(resolution =).
  • The LSI is computed internally (same results as before) instead of with Signac, which removes Signac and its genomics dependencies.
  • TF-IDF method 1 is used (chosen by grid search).
  • Clusters always get distinct labels; if needed the cluster id is appended.
  • plotDendrogram() returns a ggplot instead of drawing a base plot.
  • plotHeatmap() returns the pheatmap object invisibly and shows clusters absent from a source in gray instead of as 0.
  • plotProportions() counts a gene set found in several sources once, as “Both” or “Multiple”, and requires at least two sources.
  • Re-clustering a labeled object removes its old labels.
  • clusterGeneSets() no longer changes the random number stream of the session.

Bug fixes

  • plotUMAP(interactive = TRUE) works.
  • plotHeatmap() works with a single source and with constant values.
  • updateClusterLabels() keeps the original label when called repeatedly, validates its input, and extractTable(summarize = FALSE) reports it as initial_cluster_label.
  • plotViolin(removeClusterId = TRUE) keeps the clusters ordered by median.
  • Inputs with 50 or fewer gene sets can be clustered (at least 10 needed).
  • A resolution where the silhouette is undefined no longer stops clusterGeneSets(); if all resolutions fail, the error lists why.
  • The LSI falls back to another SVD solver when irlba does not converge.
  • Stop words with special characters, and an empty stop-word list, work.
  • Long gene-set names keep their “_UP”/“_DN” suffix when shortened.
  • plotClustree() no longer attaches ggraph to the search path.
  • Clear errors for unknown resolutions and cluster numbers.

geneSetSimplifyR 1.0.0 (2026-05)

  • Internal release, not published.
  • Updated the code to adapt to the latest version of Seurat.
  • Renamed highligthClusterNo to highlightClusterNo (and its highligthPoint* / highligthLabel* siblings) in all plotting functions.
  • Renamed plotDendogram() to plotDendrogram().
  • weighted_score in label generation weighs TF-IDF and centrality scores equally.
  • Added LazyDataCompression: xz, reducing package size.

geneSetSimplifyR 0.1.3 (2023-12)

  • Updates to fix errors introduced by a Matrix update.

geneSetSimplifyR 0.1.2 (2023-06)

  • Updated description.
  • clusterGeneSets() has new (slightly changed) default parameters based on a new grid search.
  • clusterGeneSets() detects the best clustering resolution (of the ones tested) and sets it as the default used throughout the package.
  • Plot functions handle user-supplied resolutions (plotUMAP, plotProportions, plotViolin, plotClustree) and different numbers of conditions (plotUMAP, plotViolin).
  • New plotClusterLabels() and updateClusterLabels() for in-depth analysis and fine-tuning of cluster labels.
  • plotUMAP, plotProportions and plotViolin can highlight clusters.
  • All functions use resolution as argument instead of defaultResolution.
  • Added a show method for gsList that summarizes the object.

geneSetSimplifyR 0.1.0 (2022-10-10)

  • Package established.