Node reference

Cell-Cell Communication nodes.

Infer ligand-receptor signalling between cell groups. CellPhoneDB runs the classic permutation test (run mode, or load mode for existing CLI output) with heatmap, dotplot and chord plots; LIANA adds consensus scoring (rank_aggregate over CellPhoneDB, NATMI, Connectome, logFC and more) with a ranked dotplot. For spatially aware ligand-receptor testing see the Spatial category.

04 nodesCategory · communicationin 1 template

Heads up

Oreumics bundles the CellPhoneDB code, not its database. Install a database release once through the Resource Center (an existing ~/.cpdb release or a zip you supply) — since v0.25.0 the node recognises databases installed there.

CellPhoneDB does not natively support mouse data — convert genes to HGNC human symbols first, or use LIANA with the mouseconsensus resource. LIANA’s ligand-receptor bundle ships inside the installer; nothing is fetched at run time.

LIANA expects log-normalised expression in adata.X. CellPhoneDB Plot’s cell_type1 / cell_type2 are label values, not obs column names; use . as a wildcard. Zero significant interactions is reported as a readable message (human-only, HGNC, log-normalised input) rather than a KeyError.

CellPhoneDB

cellPhoneDB

Cell-cell ligand-receptor interaction analysis. Modes: "run" (in-app) or "load" (read existing CLI output). Database setup panel handles download. v5 recommended (improved scoring). Mouse data NOT supported — convert to HGNC human symbols first.

Node LibraryscRNA-seq

InAnnData (with cell type annotation)

OutAnnData (uns["cpdb_results"])

Parameters (9)

Name	Type	Default	Description
`mode`	select	`run`	"run" or "load".
`celltype_key`	obs column	`leiden`	obs column with cell type labels.
`cpdb_file_path`	file (.zip)	—	[run mode] DB .zip path (auto-filled).
`analysis_method`	select	`statistical`	"statistical" or "basic".
`iterations`	number	`1000`	Permutation shuffles.
`threshold`	number	`0.1`	Min fraction of cells expressing a gene.
`pvalue`	number	`0.05`	P-value cutoff for downstream viz.
`threads`	number	`4`	CPU threads.
`output_dir`	directory	—	[load mode] Folder with means.txt/pvalues.txt.

CellPhoneDB Plot

cellPhoneDBViz

Visualize CellPhoneDB results — heatmap (interaction counts), dotplot (LR pairs for specific cell types), or chord (network). cell_type1/cell_type2 are LABEL VALUES (e.g. "T cell"), NOT obs column names. Use "." as wildcard.

Node LibraryscRNA-seq

InAnnData (with cpdb_results)

OutPNG image

Plots (3)heatmapdotplotchord

Parameters (9)

Name	Type	Default	Description
`plot_type`	text	`heatmap`	heatmap/dotplot/chord.
`celltype_key`	text	—	obs column (auto-detected if empty).
`cell_type1`	text	`.`	First cell type ("." = all).
`cell_type2`	text	`.`	Second cell type ("." = all).
`interaction`	text	—	[chord] Comma-separated LR pairs to highlight.
`symmetrical`	checkbox	`false`	[heatmap] Average upper/lower triangle.
`link_direction`	select	`1`	[chord] 1=ligand→, -1=receptor→, 0=undirected.
`figsize_w`	number	`10`	Figure width.
`figsize_h`	number	`8`	Figure height.

LIANA

lianaCommunication

Consensus ligand-receptor scoring (rank_aggregate) combining CellPhoneDB, NATMI, Connectome, logFC and more into one ranking of interactions between cell groups. Runs several ligand-receptor methods and aggregates them into magnitude_rank (expression strength) and specificity_rank (pair specificity). Requires log1p-normalised expression and cell type / cluster labels in obs, with gene symbols matching the chosen resource ("consensus" = human HGNC). Fetches a small L-R resource from OmniPath on first run (cached).

Node LibraryscRNA-seq

InAnnData (log1p-normalised + cell type / cluster labels, ≥2 groups)

OutAnnData (uns["liana_res"] ranked L-R table)

Parameters (5)

Name	Type	Default	Description
`celltype_key`	obs column	—	obs column with cell type / cluster labels (≥2 groups).
`resource_name`	select	`consensus`	consensus / cellphonedb / mouseconsensus / cellchatdb / connectomedb2020. "consensus" = human HGNC.
`expr_prop`	number	`0.1`	Min fraction of cells in a group expressing a gene.
`n_perms`	number	`1000`	Permutations for the specificity null distribution.
`use_raw`	select	`auto`	Expression matrix to score: auto picks the log-normalized one (adata.raw after Scale, otherwise X). Counts or scaled values are refused.

LIANA Plot

lianaVisualization

Dotplot of LIANA consensus ligand-receptor results — dot size = interaction specificity, colour = expression magnitude. Renders the top-N ranked ligand-receptor interactions across source → target cell-type pairs from a LIANA node. Filter by source/target cell types and rank by magnitude or specificity. All tuning lives in the Visualization Workspace.

Node LibraryscRNA-seq

InAnnData (uns["liana_res"] from a LIANA node)

OutPNG image

Plots (1)dotplot

Parameters (11)

Name	Type	Default	Description
`plot_type`	text	`dotplot`	Dotplot of top-ranked ligand-receptor interactions.
`top_n`	number	`20`	Number of top-ranked interactions to show.
`orderby`	select	`magnitude_rank`	"magnitude_rank" = expression strength; "specificity_rank" = pair specificity.
`source_labels`	text	—	Comma-separated source cell types. Empty = all.
`target_labels`	text	—	Comma-separated target cell types. Empty = all.
`cmap`	select	`viridis`	Colormap.
`min_significance`	number	—	Keep only interactions with -log10(specificity_rank) >= this value. Empty (null) = keep all.
`sort_by_specificity`	checkbox	`false`	Order the y-axis (L-R pairs) by specificity instead of the 'orderby' ranking.
`dot_size_range`	text	—	'min,max' dot area range (e.g. '30,230'). Empty = default 30..230.
`figsize_w`	number	`10`	Figure width in inches.
`figsize_h`	number	`8`	Figure height in inches.

Functional Activity

Visualization