functional_analysis

R functions for visualizing functional enrichment results obtained with the CERNO test from tmod package. The repository supports Gene Ontology Biological Process, Cellular Component and Molecular Function collections, as well as Reactome pathways.

CERNO_vis.R is the main script. Set the input directory, output directory and input file names in its User settings section, then run the script from the repository root.

Installation

R 4.1 or later is recommended.

source("install_packages.R")

Required input columns

CERNO results may be stored in Excel files and should contain the following columns:

Column	Meaning	Required by
ID	Pathway or gene-set identifier	tree plot, GO graph, redundancy analysis
Title	Pathway name	all visualizations
adj.P.Val	FDR-adjusted p-value	all visualizations
AUC	CERNO effect-size measure	all visualizations
ONTOLOGY	BP, CC, MF or REACTOME	added automatically by CERNO_vis.R
n_pathway_total	Pathway size	optional; used by filtering helpers

ID should preferably correspond to an MSigDB gene-set identifier. The main script normalizes IDs against the collections returned by msigdbr.

Main functions

Function	Purpose	Most important parameters
cerno_dotplot()	Displays pathways by -log10(FDR) with point size representing AUC	top_n, padj_cutoff, highlight1, highlight2, point_size_range
cerno_manhattan_go()	Compares GO terms across BP, CC and MF	padj_cutoff, label_top_n, label_padj_cutoff, max_terms_per_ontology, jitter_width
cerno_treeplot()	Hierarchically clusters pathways using Jaccard overlap of member genes	ontology, top_n, selection_mode, nCluster, cluster_method
cerno_go_graph()	Displays selected GO terms and their ancestors in the GO hierarchy	ontology, top_n, selection_mode, max_depth, include_roots, layout
cerno_simplify_go_terms()	Groups semantically redundant GO terms and selects representatives	ontology, selection_mode, similarity_cutoff, semantic_measure
cerno_go_redundancy_emap()	Visualizes semantic similarity among GO terms	label, similarity_cutoff, layout, show_removed_labels

Key parameters

The visualization functions use a consistent set of parameters wherever applicable:

• highlight1 and highlight2 are lists of exact pathway names or IDs that will be highlighted in the dotplot (by color) and treeplot (with one or two stars *, respectively), optional.

• ontology selects the functional category to analyze. Supported GO values are "BP", "CC" and "MF". The tree plot also supports "REACTOME" when appropriate pathway definitions are supplied.

• top_n defines the maximum number of terms included in the visualization.

• padj_cutoff defines the adjusted p-value threshold used to identify statistically significant terms. The default value is 0.05.

• selection_mode controls how terms are selected before constructing tree, hierarchy and redundancy visualizations:

  • "significant_top_n" selects up to top_n terms with adjusted p-values below padj_cutoff.
  • "significant_all" selects all terms with adjusted p-values below padj_cutoff.
  • "top_n" selects the top_n terms with the lowest adjusted p-values, regardless of significance.
  • "significant_or_top_n" uses significant terms when enough are available for the selected visualization; otherwise, it falls back to the top_n terms with the lowest adjusted p-values.

• ontology_col, id_col, term_col, padj_col and auc_col specify the corresponding columns in the input data. Their defaults are "ONTOLOGY", "ID", "Title", "adj.P.Val" and "AUC".

• split_after_words controls line wrapping in long pathway names.

• point_size_range controls the minimum and maximum point sizes used to represent AUC values.

• nCluster defines the number of functional clusters displayed in a pathway tree.

• similarity_cutoff defines the minimum GO semantic similarity required to group or connect terms in the redundancy analysis.

• semantic_measure selects the GO semantic similarity method. The default is "Wang".

• max_depth controls how many GO ancestor levels are included in the hierarchy graph. For greater clarity, if you have >30 significant pathways, set max_depth to 2-3.

And other parameters related to labels, legends, layout and figure dimensions.

Running the analysis

source("CERNO_vis.R")

The script creates PNG figures, CSV files describing representative and redundant GO terms in the selected output directory.

Acknowledgements

The visualization concepts and analytical workflows implemented in this repository were inspired by the functional enrichment visualization methods presented in the Biomedical Knowledge Mining book and the enrichplot ecosystem.

The code provides custom implementations adapted to tabular CERNO results.

CERNO results may be generated using the tmod R package.

References:

• Biomedical Knowledge Mining: https://yulab-smu.top/biomedical-knowledge-mining-book/enrichplot.html
• enrichplot: https://bioconductor.org/packages/enrichplot/
• tmod: https://CRAN.R-project.org/package=tmod

This repository is not affiliated with or endorsed by the authors of enrichplot, clusterProfiler or tmod.

License

This project is distributed under the GNU General Public License v3.0. See the LICENSE file for details.